Plaintiffs' Second Amended Master Complaint (Personal Injury)

Plaintiffs' filing15 Dec 2023

Narratives:Knowledge Profits over Safety

Transcribed text as recorded, including headers and formatting fragments.

Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 1 of 278 1 [Submitting Counsel on Signature Page] 2 3 4 5 6 7 8 IN THE UNITED STATES DISTRICT COURT 9 FOR THE NORTHERN DISTRICT OF CALIFORNIA 10 11 IN RE: SOCIAL MEDIA ADOLESCENT No. 4:22-MD-3047 ADDICTION/PERSONAL INJURY 12 PRODUCTS LIABILITY LITIGATION MDL No. 3047 13 14 PLAINTIFFS' SECOND This Document Relates to: AMENDED MASTER COMPLAINT 15 (PERSONAL INJURY) ALL ACTIONS 16 JURY TRIAL DEMANDED 17 18 19 20 21 22 23 24 25 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 2 of 278 1 TABLE OF CONTENTS 2 Page 3 I. INTRODUCTION .............................................................................................................. 1 4 II. THE PARTIES .................................................................................................................... 7 A. PLAINTIFFS .......................................................................................................... 7 5 B. DEFENDANTS ...................................................................................................... 8 6 1. Meta............................................................................................................. 8 7 2. Snap ............................................................................................................. 9 8 3. ByteDance ................................................................................................... 9 4. Google ....................................................................................................... 10 9 III. JURISDICTION AND VENUE ....................................................................................... 11 10 IV. FACTUAL ALLEGATIONS............................................................................................ 12 11 A. GENERAL FACTUAL ALLEGATIONS APPLICABLE TO ALL DEFENDANTS .................................................................................................... 12 12 1. Defendants have targeted children as a core market. ................................ 12 13 2. Children are uniquely susceptible to harm from Defendants' apps. ......... 14 14 3. Defendants designed their apps to attract and addict youth. ..................... 18 15 4. Millions of kids use Defendants' products compulsively. ........................ 24 16 5. Defendants' apps have created a youth mental health crisis. .................... 26 6. Defendants' defective products encourage dangerous "challenges." ....... 34 17 7. Defendants' defective social media apps facilitate and contribute to 18 the sexual exploitation and sextortion of children, and the ongoing 19 production and spread of child sex abuse material online. ....................... 37 8. Defendants could have avoided harming Plaintiffs. ................................. 45 20 9. Defendants consistently refer to and treat their apps as products. ............ 46 21 B. FACTUAL ALLEGATIONS AS TO META ....................................................... 50 22 1. Background and overview of Meta's products. ........................................ 50 23 a. Meta's origins and the development of Facebook. ....................... 52 b. Modifications of Facebook's product features over time. ............ 54 24 c. Facebook's acquisition and control of Instagram. ........................ 60 25 d. Modifications of Instagram's product features over time. ............ 63 26 2. Meta intentionally encourages youth to use its products and then 27 leverages that usage to increase revenue. .................................................. 68 3. Meta intentionally designed product features to addict children and 28 adolescents. ............................................................................................... 77 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -i- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 3 of 278 1 TABLE OF CONTENTS (continued) 2 Page 3 a. Facebook's and Instagram's algorithms maximize engagement, promoting use at levels and frequency that is 4 harmful to kids. ............................................................................. 78 5 b. Facebook's and Instagram's user interfaces are designed to create addictive engagement. ........................................................ 84 6 c. Instagram's defective product features cause negative 7 appearance comparison and social comparison ............................ 93 8 d. Meta has failed to implement effective age-verification measures to keep children off of Facebook and Instagram. ........ 104 9 e. Facebook's and Instagram's parental controls are defective. ..... 111 10 f. Facebook's and Instagram's defective features include 11 impediments to discontinuing use. .............................................. 115 4. Meta has concealed from Plaintiffs, the public, and Congress the 12 harmful effects that Instagram's and Facebook's design have on 13 children. ................................................................................................... 116 14 5. Meta facilitates the spread of CSAM and child exploitation. ................. 133 6. Meta failed to adequately warn Plaintiffs or Consortium Plaintiffs 15 about the dangers and harms caused by Instagram and Facebook, 16 or provide instructions regarding safe use. ............................................. 141 C. FACTUAL ALLEGATIONS AS TO SNAP ...................................................... 142 17 1. Background and overview of Snapchat................................................... 144 18 2. Snap targets children. .............................................................................. 145 19 a. Snap has designed its Snapchat product to grow use by 20 children to drive the company's revenue. ................................... 145 b. Snap promotes Snapchat to children. .......................................... 148 21 3. Snapchat is designed to addict children through psychological 22 manipulation............................................................................................ 152 23 a. Snap designed Snapchat to drive compulsive use through a set of social metrics and other manipulation techniques that 24 induce compulsive use. ............................................................... 152 25 b. Snap's defective features are designed to promote compulsive and excessive use. .................................................... 159 26 4. Snap designed Snapchat with features that harm children directly 27 or expose children to harm. ..................................................................... 161 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - ii - (PERSONAL INJURY) 2779607.1 CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 4 of 278 1 TABLE OF CONTENTS (continued) 2 Page 3 a. Disappearing "Snaps" and "My Eyes Only" encourage destructive behavior among Snap's teen users............................ 161 4 b. Snapchat's "Snap Map" feature endangers children. .................. 162 5 c. Snapchat's "Quick Add" feature endangers children.................. 163 6 d. Snapchat's Lenses and Filters features promote negative appearance comparison. .............................................................. 164 7 5. Snap has implemented ineffective and misleading parental 8 controls, further endangering children. ................................................... 166 9 6. Snap facilitates the spread of CSAM and child exploitation. ................. 167 10 7. Snap failed to adequately warn Plaintiffs about the harms its product causes or provide instructions regarding safe use. ..................... 171 11 D. FACTUAL ALLEGATIONS AS TO BYTEDANCE ........................................ 173 12 1. Background and overview of TikTok. .................................................... 174 13 2. ByteDance intentionally encourages youth to use its product and then leverages that use to increase revenue............................................. 175 14 3. ByteDance intentionally designed product features to addict 15 children and adolescents. ........................................................................ 176 16 a. TikTok's age-verification measures are defective. ..................... 176 b. TikTok's parental controls are defective..................................... 178 17 c. ByteDance intentionally designed TikTok's defective 18 features and algorithms to maximize engagement using automatic content, time-limited experiences, intermittent 19 variable rewards, reciprocity, and ephemeral content. ................ 179 20 d. ByteDance's defective features include impediments to 21 discontinuing use......................................................................... 192 e. ByteDance's defective features inflict impossible image 22 standards and encourage negative appearance comparison. ....... 193 23 4. ByteDance facilitates the spread of CSAM and child exploitation......... 195 24 5. ByteDance failed to adequately warn Plaintiffs about the harms its product causes or to provide instructions regarding safe use. ................. 198 25 E. FACTUAL ALLEGATIONS AS TO GOOGLE ................................................ 202 26 1. Background and overview of YouTube. ................................................. 202 27 2. Google intentionally encourages youth to use YouTube and then leverages that use to increase revenue. ................................................... 204 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - iii - (PERSONAL INJURY) 2779607.1 CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 5 of 278 1 TABLE OF CONTENTS (continued) 2 Page 3 3. Google intentionally designed product features to addict children and adolescents........................................................................................ 209 4 a. Google's age-verification measures and parental controls 5 are defective. ............................................................................... 209 6 b. YouTube is defectively designed to inundate users with features that use intermittent variable rewards and 7 reciprocity. .................................................................................. 211 8 c. Google's algorithms are designed to maximize "watch time." ........................................................................................... 215 9 d. YouTube's defective features include impediments to 10 discontinuing use......................................................................... 225 11 4. Google facilitates the spread of CSAM and child exploitation............... 226 5. Google failed to adequately warn Plaintiffs about the harm its 12 products cause or provide instructions regarding safe use. ..................... 232 13 V. TIMELINESS AND TOLLING OF STATUTES OF LIMITATIONS ......................... 233 14 VI. PLAINTIFFS' CLAIMS ................................................................................................. 234 COUNT 1: STRICT LIABILITY – DESIGN DEFECT (Against All Defendants) ...... 235 15 COUNT 2: STRICT LIABILITY – FAILURE TO WARN (Against All 16 Defendants) ......................................................................................................... 240 17 COUNT 3: NEGLIGENCE – DESIGN (Against All Defendants) ............................... 243 18 COUNT 4: NEGLIGENCE – FAILURE TO WARN (Against All Defendants).......... 246 COUNT 5: NEGLIGENCE (Against All Defendants) ................................................... 249 19 COUNT 6: INTENTIONALLY OMITTED................................................................... 254 20 COUNT 7: VIOLATION OF UNFAIR TRADE PRACTICES/CONSUMER 21 PROTECTION LAWS (Against All Defendants) ............................................. 254 22 COUNT 8: FRAUDULENT CONCEALMENT AND MISREPRESENTATION (Against Meta) .................................................................................................... 257 23 COUNT 9: NEGLIGENT CONCEALMENT AND MISREPRESENTATION 24 (Against Meta) .................................................................................................... 259 COUNT 10: NEGLIGENCE PER SE (Against All Defendants) .................................. 261 25 COUNT 11: INTENTIONALLY OMITTED................................................................. 263 26 COUNT 12: VIOLATIONS OF 18 U.S.C. §§ 2255 and 2252 (Against Meta) ............. 264 27 COUNT 13: INTENTIONALLY OMITTED................................................................. 265 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - iv - (PERSONAL INJURY) 2779607.1 CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 6 of 278 1 TABLE OF CONTENTS (continued) 2 Page 3 COUNT 14: VIOLATIONS OF 18 U.S.C. §§ 2255, and 2252A(5)(b) (Against Meta) ................................................................................................................... 265 4 COUNT 15: INTENTIONALLY OMITTED................................................................. 267 5 COUNT 16: WRONGFUL DEATH (Against All Defendants) .................................... 267 6 COUNT 17: SURVIVAL ACTION (Against All Defendants) ..................................... 267 7 COUNT 18: LOSS OF CONSORTIUM AND SOCIETY (Against All Defendants) ......................................................................................................... 268 8 VII. PRAYER FOR RELIEF .................................................................................................. 269 9 VIII. JURY DEMAND ............................................................................................................ 270 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -v- (PERSONAL INJURY) 2779607.1 CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 7 of 278 1 I. INTRODUCTION 2 1. American children are suffering an unprecedented mental health crisis fueled by 3 Defendants' addictive and dangerous social media products. 4 2. In the past decade, Americans' engagement with social media grew exponentially, 5 nowhere more dramatically than among our country's youth. That explosion in usage is no accident. 6 It is the result of Defendants' studied efforts to induce young people to compulsively use their 7 products—Instagram, Facebook, TikTok, Snapchat, and YouTube. Borrowing heavily from the 8 behavioral and neurobiological techniques used by slot machines and exploited by the cigarette 9 industry, Defendants deliberately embedded in their products an array of design features aimed at 10 maximizing youth engagement to drive advertising revenue. Defendants know children are in a 11 developmental stage that leaves them particularly vulnerable to the addictive effects of these 12 features. Defendants target them anyway, in pursuit of additional profit. 13 3. The defects in Defendants' products vary by platform, but all exploit children and 14 adolescents. They include but are not limited to an algorithmically-generated, endless feed to keep 15 users scrolling in an induced "flow state;" "intermittent variable rewards" that manipulate 16 dopamine delivery to intensify use; "trophies" to reward extreme usage; metrics and graphics to 17 exploit social comparison; incessant notifications that encourage repetitive account checking by 18 manufacturing insecurity; inadequate, essentially illusory age verification protocols; and deficient 19 tools for parents that create the illusion of control. 20 4. The resulting ubiquity of Defendants' products in the lives and palms of our kids, 21 and the ensuing harm to them, is hard to overstate. Today, over a third of 13- to 17-year-old kids 22 report using one of Defendants' apps "almost constantly" and admit this is "too much." Yet more 23 than half of these kids report that they would struggle to cut back on their social media use. Instead 24 of feeding coins into machines, kids are feeding Defendants' platforms with an endless supply of 25 attention, time, and data. 26 5. Defendants' choices have generated extraordinary corporate profits—and yielded 27 immense tragedy. Suicide rates for youth are up an alarming 57%. Emergency room visits for 28 anxiety disorders are up 117%. In the decade leading up to 2020, there was a 40% increase in high PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -1- (PERSONAL INJURY) 2779607.1 CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 8 of 278 1 school students reporting persistent sadness and hopelessness, and a 36% increase in those who 2 attempted to take their own lives. In 2019, one in five high school girls had made a suicide plan. In 3 2021, one in three girls seriously considered attempting suicide. Children and their parents and 4 guardians across the country have struggled to cope with the severe, lasting damage visited on their 5 families by anxiety, depression, addiction, eating disorders, self-harm, suicidality, and the loss of 6 outliving one's child. 7 6. This lawsuit follows on a growing body of scientific research, including Defendants' 8 own internal (previously concealed) studies, that draws a direct line between Defendants' 9 conscious, intentional design choices and the youth mental health crisis gripping our nation. 10 Instagram, Facebook, TikTok, Snapchat, and YouTube have rewired how our kids think, feel, and 11 behave. Disconnected "Likes" have replaced the intimacy of adolescent friendships. Mindless 12 scrolling has displaced the creativity of play and sport. While presented as "social," Defendants' 13 products have in myriad ways promoted disconnection, disassociation, and a legion of resulting 14 mental and physical harms. 15 7. The U.S. Surgeon General recently explained that children versus Big Tech is "just 16 not a fair fight."1 "You have some of the best designers and product developers in the world who 17 have designed these products to make sure people are maximizing the amount of time they spend 18 on these platforms. And if we tell a child, use the force of your willpower to control how much 19 time you're spending, you're pitting a child against the world's greatest product designers." 20 8. Over the past year, hundreds of personal injury actions have been filed in courts 21 across the country alleging that Defendants defectively designed their platforms—in foreseeably 22 unsafe ways and in dereliction of their basic duties of care—to induce harmful, unhealthy, and 23 compulsive use by kids. Plaintiffs in these cases are the young people across the country whose 24 descent into the void of social media has led to serious and sometimes fatal harm, and their parents 25 26 1 27 Allison Gordon & Pamela Brown, Surgeon General says 13 is 'too early' to join social media, CNN (Jan. 29, 2023), https://www.cnn.com/2023/01/29/health/surgeon-general-social- 28 media/index.html. Exhibits and referenced materials are incorporated in this Master Complaint as if fully stated herein. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -2- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 9 of 278 1 and guardians. Defendants are the multibillion-dollar corporations who designed unsafe products 2 that hopelessly outmatch parents' struggle to keep their children healthy and safe. 3 9. Plaintiffs file this Master Complaint (Personal Injury) ("Complaint") as an 4 administrative device, to set forth the potential claims and facts that individual Plaintiffs may assert 5 in this multidistrict proceeding against Defendants. 2 Unless otherwise indicated, Plaintiffs make 6 allegations about themselves based on personal knowledge, and allegations about Defendants on 7 information and belief generally gained through their attorneys' investigations. 8 *** 9 10. Over the past decade, Defendants have relentlessly pursued a strategy of growth-at- 10 all-costs, recklessly ignoring the impact of their products on children's mental and physical health 11 and well-being.3 In a race to corner the "valuable but untapped" market of tween and teen users, 12 each Defendant designed product features to promote repetitive, uncontrollable use by kids. 4 13 11. Adolescents and children are central to the Defendants' business models. These age 14 groups are highly connected to the Internet, more likely to have social media accounts, and more 15 likely to devote their downtime to social media usage. Additionally, youth influence the behavior 16 of their parents and younger siblings. As one Defendant put it, "los[ing] the teen foothold in the 17 U.S." would mean "los[ing] the pipeline" for growth. 5 18 19 2 Case Management Order No. 3, Dkt. No. 111 at 2; see In re Propulsid Products Liab. Litig., 208 20 F.R.D. 133, 141 (E.D. La. 2002). The Master Complaint does not necessarily include all claims or allegations that have been or will be asserted in each action filed in, or transferred to, this Court. 21 Individual plaintiffs may adopt the allegations and claims in this Master Complaint through a separate Short Form Complaint. Individual plaintiffs may supplement or add allegations, claims, 22 or defendants to their respective Short Form Complaints. This Master Complaint does not waive or dismiss any claims in any individual action. Nor does any Plaintiff relinquish any right they 23 otherwise would have had, absent this Master Complaint, to amend (or move to amend) their Short Form Complaints. 24 3 See, e.g., Haugen_00000934 (admission by a Software Engineer at Meta, that "It's not a secret that we've often resorted to aggressive tactics in the name of growth, and we've been pretty 25 unapologetic about it."). 4 Georgia Wells & Jeff Horwitz, Facebook's Effort to Attract Preteens Goes Beyond Instagram 26 Kids, Documents Show, Wall St. J. (Sept. 28, 2021), https://www.wsj.com/articles/facebook- instagram-kids-tweens-attract-11632849667; see also Haugen_00022339. 27 5 Sheera Frenkel et al., Instagram Struggles with Fears of Losing Its 'Pipeline': Young Users, 28 N.Y. Times (Oct. 26, 2021), available at https://www.nytimes.com/2021/10/16/technology/instagram-teens.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -3- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 10 of 278 1 12. Recognizing the power of engaging young users, Defendants deliberately tweaked 2 the design and operation of their apps to exploit the psychology and neurophysiology of kids. 3 Because children's and adolescents' brains are not fully developed, they lack the same emotional 4 maturity, impulse control, and psychological resiliency as adults. As a result, they are uniquely 5 susceptible to addictive features in digital products and highly vulnerable to the consequent harms. 6 Knowing this, Defendants wrote code designed to manipulate dopamine release in children's 7 developing brains and, in doing so, create compulsive use of their apps. 8 13. Defendants' strategy paid off. Users of their products now number in the billions, 9 and the frequency and time spent by these users has grown exponentially. This has allowed 10 Defendants to harvest a vast amount of personal user data—from the school you attend, to the 11 sneakers you covet, to the places you've been and the people you've met. This, in turn, has allowed 12 Defendants to mint a fortune, by selling to others the ability to micro-target advertisements to 13 incredibly narrow slices of the public.6 14 14. Defendants' growth has come at the expense of its most vulnerable users: children 15 around the world, including Plaintiffs, who they cultivated and exploited. Plaintiffs are not merely 16 the collateral damage of Defendants' products. They are the direct victims of the intentional product 17 design choices made by each Defendant. They are the intended targets of the harmful features that 18 pushed them into self-destructive feedback loops. 19 15. As a direct result of Defendants' successful promotion of their defective products, 20 the rates of mental health issues among children have climbed steadily since 2010. By 2018, suicide 21 was the second leading cause of death for youth. 7 22 23 24 25 6 See Snap, Inc., 2022 Annual Report (Form 10-K) at 15 (Jan. 31, 2023) ("[W]e rely heavily on our ability to collect and disclose data[] and metrics to our advertisers so we can attract new 26 advertisers and retain existing advertisers. Any restriction or inability, whether by law, regulation, policy, or other reason, to collect and disclose data and metrics which our advertisers find useful 27 would impede our ability to attract and retain advertisers."). 7 28 CDC, Deaths: Leading Causes for 2018, 70(4) National Vital Statistics Reports at 10 (May 17, 2021), https://www.cdc.gov/nchs/data/nvsr/nvsr70/nvsr70-04-508.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -4- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 11 of 278 1 16. In late 2021, the U.S. Surgeon General issued an advisory "to highlight the urgent 2 need to address the nation's youth mental health crisis."8 In a scathing rebuke of the assault on our 3 children, the Surgeon General recognized the dangerous designs in Defendants' products and 4 Defendants' abdication of responsibility for the resulting harms: 5 In these digital public spaces, which are privately owned and tend to be run for profit, there can be tension between what's best for the 6 technology company and what's best for the individual user or for society. Business models are often built around maximizing user 7 engagement as opposed to safeguarding users' health and ensuring that users engage with one another in safe and healthy ways . . . . 8 [T]echnology companies must step up and take responsibility for 9 creating a safe digital environment for children and youth. Today, most companies are not transparent about the impact of their 10 products, which prevents parents and young people from making informed decisions and researchers from identifying problems and 11 solutions.9 12 17. The Surgeon General's comments have since been echoed by President Biden 13 himself. In both his 2022 and 2023 State of the Union Addresses, the President urged the nation to 14 "hold social media platforms accountable for the national experiment they're conducting on our 15 children for profit."10 In a January 11, 2023 op-ed, President Biden amplified this point: "The risks 16 Big Tech poses for ordinary Americans are clear. Big Tech companies collect huge amounts of data 17 on the things we buy, on the websites we visit, on the places we go and, most troubling of all, on 18 our children."11 The President observed that "millions of young people are struggling with bullying, 19 20 21 8 Press Release, U.S. Dep't Health & Hum. Servs., U.S. Surgeon General Issues Advisory on 22 Youth Mental Health Crisis Further Exposed by COVID-19 Pandemic (Dec. 7, 2021), https://www.hhs.gov/about/news/2021/12/07/us-surgeon-general-issues-advisory-on-youth- 23 mental-health-crisis-further-exposed-by-covid-19-pandemic.html. 9 U.S. Surgeon General's Advisory, Protecting Youth Mental Health (Dec. 7, 2021), 24 https://www.hhs.gov/sites/default/files/surgeon-general-youth-mental-health-advisory.pdf (emphasis in original). 25 10 The White House, President Biden's State of the Union Address (Mar. 1, 2022), https://www.whitehouse.gov/state-of-the-union-2022/; see also The White House, President 26 Biden's State of the Union Address (Feb. 7, 2023), https://www.whitehouse.gov/state-of-the- union-2023/. 27 11 Joe Biden, Republicans and Democrats, Unite Against Big Tech Abuses, Wall St. J. (Jan. 11, 28 2023), https://www.wsj.com/articles/unite-against-big-tech-abuses-social-media-privacy- competition-antitrust-children-algorithm-11673439411. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -5- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 12 of 278 1 violence, trauma and mental health" as a result of Defendants' conduct and products, and again 2 stated that "[w]e must hold social-media companies accountable" for their role in this crisis. 12 3 18. These statements by President Biden and the Surgeon General are in line with a 4 substantial body of peer-reviewed scientific literature documenting the harmful impact that 5 Defendants' apps have on our children, including the various injuries suffered by Plaintiffs. This 6 body of research demonstrates that Defendants' defectively designed products can cause the harms 7 Plaintiffs suffer: addiction, compulsive use, anxiety, depression, eating disorders, body 8 dysmorphia, self-harm, sexual exploitation, suicidal ideations, other serious diseases and injuries, 9 and suicide itself. Overall rates of these disorders have increased greatly because of widespread 10 consumption of Defendants' products by children in this country and across the world. 11 19. Defendants knew or should have known about the risks of such addiction—which 12 at least one Defendant euphemistically calls "problematic use." 13 They could have changed their 13 products to avoid the harm. They could have warned the public and Plaintiffs about the danger. 14 Instead, Defendants placed growth first. 15 20. Plaintiffs seek to recover damages from Defendants and hold them responsible for 16 personal injuries resulting from their wrongful conduct. That conduct includes: (a) designing 17 defective products that caused serious injuries to users; (b) failing to provide adequate warnings 18 about serious and reasonably foreseeable health risks from use of the products; (c) failing to utilize 19 reasonable care in, among other things, developing, designing, managing, operating, testing, 20 producing, labeling, marketing, advertising, promoting, controlling, selling, supplying, and 21 distributing their products; and (d) as to Meta, engaging in the deliberate concealment, 22 23 12 Joe Biden, Republicans and Democrats, Unite Against Big Tech Abuses, Wall St. J. (Jan. 11, 2023), https://www.wsj.com/articles/unite-against-big-tech-abuses-social-media-privacy- 24 competition-antitrust-children-algorithm-11673439411. 13 25 See Haugen_00016373 at Haugen_00016379 (internal Meta report from March 2020 summarizing internal research on "problematic use"—when a user "experienc[es] both of the 26 following issues 'very often' or 'all the time': Lack of control or feelings of guilt over Facebook use. Negative impact in at least one of the following areas: productivity, sleep, parenting, or 27 relationships.");Haugen_00016373 at Haugen_00016412, Haugen_00016490 (referring to "problematic use" as "Loss of Control Over Time Spent" or "LCOTS"); Haugen_00016373 at 28 Haugen_00016379 (recognizing that "Problematic Use" is "sometimes referred to as 'social media addiction' externally"). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -6- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 13 of 278 1 misrepresentation, and obstruction of public awareness of serious health risks to users of its 2 products. 3 II. THE PARTIES 4 A. PLAINTIFFS 5 21. This Complaint is filed on behalf of children who suffered personal injuries—and, 6 in cases of death, the personal representatives of their estates ("Plaintiffs")—due to their use of 7 Defendants' products and, where applicable, their parents, guardians, spouses, children, siblings, 8 and close family members, who suffered loss of society or consortium and other injuries as a 9 consequence of the harms to Plaintiffs ("Consortium Plaintiffs"), who file a Short Form Complaint. 10 By operation of an anticipated Court order, all allegations pled in this Complaint are deemed pled 11 in any Short Form Complaint as to the Defendants identified therein. 12 22. Plaintiffs have suffered various personal injuries because of their use of Defendants' 13 products. Plaintiffs and Consortium Plaintiffs have been harmed as a direct and proximate result of 14 Defendants' wrongful conduct. These harms include pain, suffering, disability, impairment, 15 disfigurement, death, an increased risk of injury and other serious illnesses, loss of enjoyment of 16 life, loss of society, aggravation or activation of preexisting conditions, scarring, inconvenience, 17 incurred costs for medical care and treatment, loss of wages and wage-earning capacity, and other 18 economic and non-economic damages (specifically including any injuries set forth in a Short Form 19 Complaint). These losses are often permanent and continuing in nature. 20 23. Plaintiffs expressly disaffirm any contract they may have made with any of the 21 Defendants, or that Defendants may claim they made with them, before reaching the age of 22 majority, as they lacked capacity to contract. 23 24. Plaintiffs also expressly disaffirm any contract they may have made with any of the 24 Defendants, or that Defendants may claim they made with them, after reaching the age of majority, 25 because Plaintiffs' continued use of Defendants' products was compulsive and due to addiction, 26 not an affirmation of any contract. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -7- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 14 of 278 1 B. DEFENDANTS 2 25. The defendants identified in this section are collectively referred to as "Defendants" 3 throughout this Complaint. 4 1. Meta 5 26. Defendant Meta Platforms, Inc. ("Meta Platforms") is a Delaware corporation and 6 multinational technology conglomerate. Its principal place of business is in Menlo Park, CA. 7 27. Meta Platforms' subsidiaries include, but may not be limited to, the entities 8 identified in this section, as well as a dozen others whose identity or involvement is presently 9 unclear. 10 28. Defendant Facebook Payments, Inc. ("Facebook 1") is a wholly owned subsidiary 11 of Meta Platforms that was incorporated in Florida on December 10, 2010. Facebook 1 manages, 12 secures, and processes payments made through Meta Platforms, among other activities. Its principal 13 place of business is in Menlo Park, CA. 14 29. Defendant Siculus, Inc. ("Siculus") is a wholly owned subsidiary of Meta Platforms 15 that was incorporated in Delaware on October 19, 2011. Siculus constructs data facilities to support 16 Meta Platforms' products. Its principal place of business is in Menlo Park, CA. 17 30. Defendant Facebook Operations, LLC ("Facebook 2") is a wholly owned subsidiary 18 of Meta Platforms that was incorporated in Delaware on January 8, 2012. Facebook 2 is likely a 19 managing entity for Meta Platforms' other subsidiaries. Meta Platforms is the sole member of this 20 LLC, whose principal place of business is in Menlo Park, CA. 21 31. Defendant Instagram, LLC ("Instagram, LLC") launched an app called Instagram in 22 October 2010. On or around April 7, 2012, Meta Platforms purchased Instagram, LLC for over one 23 billion dollars and reincorporated the company in Delaware. Meta Platforms is the sole member of 24 this LLC, whose principal place of business is in Menlo Park, CA. 25 32. Meta Platforms, Instagram, Siculus, Facebook 1, and Facebook 2 are referred to 26 jointly as "Meta." 27 33. Meta owns, operates, controls, produces, designs, maintains, manages, develops, 28 tests, labels, markets, advertises, promotes, supplies, and distributes digital products available PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -8- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 15 of 278 1 through mobile- and web-based applications ("apps"), including Instagram and Facebook (together, 2 "Meta products"); Messenger; and Messenger Kids. Meta's apps and devices are widely distributed 3 to consumers throughout the United States. 4 2. Snap 5 34. Defendant Snap Inc. ("Snap") is a Delaware corporation. Its principal place of 6 business is in Santa Monica, CA. 7 35. Snap owns, operates, controls, produces, designs, maintains, manages, develops, 8 tests, labels, markets, advertises, promotes, supplies, and distributes the app Snapchat. Snapchat is 9 widely available to consumers throughout the United States. 10 3. ByteDance 11 36. Defendant ByteDance Ltd. is a global company incorporated in the Cayman Islands. 12 Its principal place of business is in Beijing, China. ByteDance Ltd. also maintains offices in the 13 United States, Singapore, India, and the United Kingdom, among other locations. 14 37. ByteDance Ltd. wholly owns its subsidiary Defendant ByteDance Inc., a Delaware 15 corporation whose principal place of business is in Mountain View, CA. 16 38. ByteDance Ltd.'s key Chinese subsidiary is Beijing Douyin Information Service 17 Limited, f/k/a Beijing ByteDance Technology Co. Ltd. ("Beijing ByteDance"). 14 Beijing 18 ByteDance owns, operates, and holds key licenses to Douyin, the Chinese version of TikTok. On 19 or around April 30, 2021, the Chinese government took a 1% stake in, and received one of three 20 seats on the board of directors of, Beijing ByteDance. 15 Specifically, 1% of Beijing ByteDance is 21 now owned by WangTouZhongWen (Beijing) Technology, which in turn is owned by China 22 Internet Investment Fund (China's top Internet regulator and censor), China Media Group (China's 23 24 14 See Sophie Webster, ByteDance Changes Names of Subsidiaries to Douyin, Speculated to be 25 Mulling an IPO, Tech Times (May 8, 2022), available at https://www.techtimes.com/articles/275188/20220508/bytedance-changes-names-subsidiaries- 26 douyin-speculated-mulling-ipo.htm. 15 27 See Juro Osawa & Shai Oster, Beijing Tightens Grip on ByteDance by Quietly Taking Stake, China Board Seat, The Information (Aug. 16, 2021), available at 28 https://www.theinformation.com/articles/beijing-tightens-grip-on-bytedance-by-quietly-taking- stake-china-board-seat?rc=ubpjcg. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -9- (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 16 of 278 1 national broadcaster, controlled by the Chinese Communist Party's propaganda department), and 2 the Beijing municipal government's investment arm. 3 39. ByteDance Ltd. wholly owns its subsidiary Defendant TikTok, Ltd., a Cayman 4 Island corporation with its principal place of business in Shanghai, China. 5 40. TikTok, Ltd. wholly owns its subsidiary Defendant TikTok, LLC which is, and at 6 all relevant times was, a Delaware limited liability company. 7 41. TikTok, LLC wholly owns its subsidiary Defendant TikTok, Inc. f/k/a Musical.ly, 8 Inc. ("TikTok, Inc."), a California corporation with its principal place of business in Culver City, 9 CA. 10 42. Defendants TikTok, Ltd.; TikTok, LLC; TikTok, Inc.; ByteDance Ltd.; and 11 ByteDance Inc. are referred to jointly as "ByteDance." 12 43. ByteDance owns, operates, controls, produces, designs, maintains, manages, 13 develops, tests, labels, markets, advertises, promotes, supplies, and distributes the app TikTok. 14 TikTok is widely available to consumers throughout the United States. 15 4. Google 16 44. Google Inc. was incorporated in California in September 1998 and reincorporated 17 in Delaware in August 2003. In or around 2017, Google Inc. converted to a Delaware limited 18 liability company, Defendant Google, LLC (together with its predecessor-in-interest Google Inc., 19 "Google"). Google's principal place of business is in Mountain View, CA. 20 45. Since 2006, Google has operated, done business as, and wholly owned as its 21 subsidiary Defendant YouTube, LLC ("YouTube, LLC"). YouTube, LLC is a Delaware limited 22 liability company with its principal place of business in San Bruno, CA. YouTube is widely 23 available to consumers throughout the United States.16 24 46. On October 2, 2015, Google reorganized and became a wholly owned subsidiary of 25 a new holding company, Alphabet Inc., a Delaware corporation with its principal place of business 26 in Mountain View, CA. 27 16 See, e.g., Alphabet Inc., Form 10-Q, Oct. 25, 2022, at 4 (defining Alphabet as "Alphabet Inc. 28 and its subsidiaries."), available at https://www.sec.gov/Archives/edgar/data/1652044/000165204422000090/goog-20220930.htm. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 10 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 17 of 278 1 47. Google, LLC and YouTube, LLC (together, "Google") are alter egos of one another: 2 together and in concert they own, operate, control, produce, design, maintain, manage, develop, 3 test, label, market, advertise, promote, supply, and distribute the app YouTube. 4 III. JURISDICTION AND VENUE 5 48. This Court has subject-matter jurisdiction over various constituent cases in this 6 multidistrict litigation pursuant to 28 U.S.C. § 1332(a) because Plaintiffs and Defendants in such 7 cases are residents of different states, and the amount in controversy exceeds $75,000. 8 49. This Court has subject-matter jurisdiction over various constituent cases in this 9 multidistrict litigation pursuant to 28 U.S.C. § 1331 because they involve questions of federal law 10 arising under 18 U.S.C. §§ 2255, 1595, 2252A(f), and/or 2258B. This Court has supplemental 11 jurisdiction under 28 U.S.C. § 1367(a) over Plaintiffs' state law claims because all claims alleged 12 herein form part of the same case or controversy. 13 50. This Court has personal jurisdiction over Defendants because they are incorporated 14 in and have their principal places of business in California, and because they have contacts with 15 California that are so continuous and systematic that they are essentially at home in this state. Meta, 16 Google, and ByteDance, Inc. maintain their principal places of business within this District. Snap 17 and TikTok Inc. maintain their headquarters in this State. All Defendants regularly conduct and 18 solicit business in California, provide products and/or services by or to persons here, and derive 19 substantial revenue from the same. All Defendants affirmatively and extensively engage with a 20 significant percentage of this State's residents through messages, notifications, recommendations, 21 and other communications. 22 51. This Court also has personal jurisdiction over Defendants because they committed 23 the acts complained of herein in this State and District and, in some cases, caused injury to persons 24 located in this State and District. 25 52. Venue is proper in this multidistrict litigation for one or more of the following 26 reasons: (1) at least one Defendant is headquartered and subject to personal jurisdiction in this 27 District and all Defendants are residents of this State; (2) a substantial part of the events and 28 omissions giving rise to Plaintiffs' causes of action occurred in this District; (3) venue is proper in PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 11 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 18 of 278 1 the transferor court in each constituent case; and/or (4) venue is proper in this District for pretrial 2 purposes pursuant to 28 U.S.C. § 1407 and the October 6, 2022 Transfer Order of the Judicial Panel 3 on Multidistrict Litigation. 4 53. Declaratory relief is proper under 28 U.S.C. §§ 2201 and 2202. 5 IV. FACTUAL ALLEGATIONS 6 A. GENERAL FACTUAL ALLEGATIONS APPLICABLE TO ALL DEFENDANTS 7 1. Defendants have targeted children as a core market. 8 9 54. Each Defendant has designed, engineered, marketed, and operated its products to 10 maximize the number of children who download and use them compulsively. Children are more 11 vulnerable users and have more free time on their hands than their adult counterparts. Because 12 children use Defendants' products more, they see more ads, and as a result generate more ad 13 revenue for Defendants. Young users also generate a trove of data about their preferences, habits, 14 and behaviors. That information is Defendants' most valuable commodity. Defendants mine and 15 commodify that data, including by selling to advertisers the ability to reach incredibly narrow 16 tranches of the population, including children. Each Defendant placed its app(s) into the stream of 17 commerce and generated revenues through the distribution of those apps at the expense of the 18 consuming public and Plaintiffs. 19 55. This exploitation of children, including each of the individual Plaintiffs in these 20 actions, has become central to Defendants' profitability. Like the cigarette industry a generation 21 earlier, Defendants understand that a child user today becomes an adult user tomorrow. 17 Indeed, 22 Defendants' insatiable appetite for growth has created a need for younger and younger users. 23 Defendants' wrongfully acquired knowledge of their childhood userbase has allowed them to 24 develop product designs to target elementary school-age children, who are uniquely vulnerable. 25 17 26 Haugen_00006240 ("There are many lines of evidence for a substantial 'ratchet' effect in the growth of social apps: once you get a user on your app it's hard to lose them. More precisely: the 27 adoption of an app at a given point in time depends not just on the features of that app today, but is [sic] also depends on the previous adoption of that app."); id. at Haugen_00006241 (noting 28 that, because of sunk costs and network effect, users will "stick with [an app] even if the relative quality declines."). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 12 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 19 of 278 1 Like Joe Camel of old, Defendants' recent attempts to capture pre-adolescent audiences include 2 "kid versions" of apps that are "designed to fuel [kids'] interest in the grown-up version." 18 3 56. Recognizing the vulnerability of children under 13, particularly in the Internet age, 4 Congress enacted the Children's Online Privacy Protection Act ("COPPA") in 1999. 19 COPPA 5 regulates the conditions under which Defendants can collect, use, or disclose the personal 6 information of children under 13. Under COPPA, developers of apps and websites that are directed 7 to or known to be used by children under 13 cannot lawfully obtain the individually identifiable 8 information of such children without first obtaining verifiable consent from their parents. 20 Even 9 apart from COPPA, it is well established under the law that children lack the legal or mental 10 capacity to make informed decisions about their own well-being. 11 57. COPPA was enacted precisely because Congress recognized that children under age 12 13 are particularly vulnerable to being taken advantage of by unscrupulous website operators. As a 13 June 1998 report by the FTC observed, "[t]he immediacy and ease with which personal information 14 can be collected from children online, combined with the limited capacity of children to understand 15 fully the potentially serious safety and privacy implications of providing that information, have 16 created deep concerns about current information practices involving children online." 21 The same 17 report observed that children under the age of 13 "generally lack the developmental capacity and 18 judgment to give meaningful consent to the release of personal information to a third party." 22 19 18 Leonard Sax, Is TikTok Dangerous for Teens?, Inst. Fam. Stud. (Mar. 29, 2022), 20 https://ifstudies.org/blog/is-tiktok-dangerous-for-teens-. 19 21 See 15 U.S.C. §§ 6501-6506. 20 The FTC recently clarified that acceptable methods for obtaining verifiable parent consent 22 include: (a) providing a form for parents to sign and return; (b) requiring the use of a credit/card online payment that provides notification of each transaction; (c) connecting to trained personnel 23 via video conference; (d) calling a staffed toll-free number; (e) asking knowledge-based questions; or (f) verifying a photo-ID from the parent compared to a second photo using facial 24 recognition technology. Federal Trade Commission, Complying with COPPA: Frequently Asked Questions, July 2020, https://www.ftc.gov/business-guidance/resources/complying-coppa- 25 frequently-asked-questions. 21 Privacy Online: A Report to Congress, Federal Trade Commission (1998) at 13. 26 https://www.ftc.gov/sites/default/files/documents/reports/privacy-online-report-congress/priv- 23a.pdf. 27 22 Privacy Online: A Report to Congress, Federal Trade Commission (1998) at 13. 28 https://www.ftc.gov/sites/default/files/documents/reports/privacy-online-report-congress/priv- 23a.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 13 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 20 of 278 1 58. Contemporaneous testimony by the Chairman of the FTC observed that the Internet 2 "make[s] it easy for children to disclose personal information to the general public without their 3 parents' awareness or consent. Such public disclosures raise safety concerns." 23 Further, "the 4 practice of collecting personal identifying information directly from children without parental 5 consent is clearly troubling, since it teaches children to reveal their personal information to 6 strangers and circumvents parental control over their family's information." 24 7 59. None of the Defendants conduct proper age verification or authentication. Instead, 8 each Defendant leaves it to users to self-report their age. This unenforceable and facially inadequate 9 system allows children under 13 to easily create accounts on Defendants' apps. 10 60. This is particularly egregious for two reasons. First, Defendants have long been on 11 notice of the problem. For instance, in May 2011, Consumer Reports reported the "troubling news" 12 that 7.5 million children under 13 were on Facebook. 25 Second, given that Defendants have 13 developed and utilized age-estimation algorithms for the purpose of selling user data and targeted 14 advertisements, Defendants could readily use these algorithms to prevent children under 13 from 15 accessing their products, but choose not to do so. Instead, they have turned a blind eye to collecting 16 children's data in violation of COPPA. 17 61. Defendants have done this because children are financially lucrative, particularly 18 when they are addicted to Defendants' apps. 19 2. Children are uniquely susceptible to harm from Defendants' apps. 20 62. Young people are not only Defendants' most lucrative market, but are also those 21 most vulnerable to harms resulting from Defendants' products. 22 23 24 23 S. 2326, Children's Online Privacy Protection Act of 1998: Hearing Before the U.S. Sen. 25 Subcom. On Communications, Comm. On Commerce, Science, and Transportation, 105 th Cong. 11 (1998) (statement of Robert Pitofsky, Chairman, Federal Trade Commission). 26 24 S. 2326, Children's Online Privacy Protection Act of 1998: Hearing Before the U.S. Sen. Subcom. On Communications, Comm. On Commerce, Science, and Transportation, 105 th Cong. 27 11 (1998) (statement of Robert Pitofsky, Chairman, Federal Trade Commission). 25 28 Emily Bazelon, Why Facebook is After Your Kids, N.Y. Times (Oct. 12, 2011), https://www.nytimes.com/2011/10/16/magazine/why-facebook-is-after-your-kids.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 14 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 21 of 278 1 63. "Everyone innately responds to social approval." 26 "[B]ut some demographics, in 2 particular teenagers, are more vulnerable to it than others."27 Unlike adults, who "tend to have a 3 fixed sense of self that relies less on feedback from peers," 28 adolescents are in a "period of personal 4 and social identity formation" that "is now reliant on social media." 29 5 64. To understand the impact Defendants' apps have on young people, it is helpful to 6 understand some basics about the human brain. 7 65. The frontal lobes of the brain—particularly the prefrontal cortex—control higher- 8 order cognitive functions. This region of the brain is central to planning and executive decision- 9 making, including the evaluation of risk and reward. It also helps inhibit impulsive actions and 10 "regulate emotional responses to social rewards." 30 11 66. Children and adolescents are especially vulnerable to developing harmful behaviors 12 because their prefrontal cortex is not fully developed.31 Indeed, it is one of the last regions of the 13 brain to mature.32 In the images below, the blue color depicts brain development. 33 14 15 26 Von Tristan Harris, The Slot Machine in Your Pocket, Spiegel Int'l (July 27, 2016), 16 https://www.spiegel.de/international/zeitgeist/smartphone-addiction-is-part-of-the-design-a- 1104237.html. 27 17 Von Tristan Harris, The Slot Machine in Your Pocket, Spiegel Int'l (July 27, 2016), https://www.spiegel.de/international/zeitgeist/smartphone-addiction-is-part-of-the-design-a- 18 1104237.html. 28 Zara Abrams, Why young brains are especially vulnerable to social media, Am. Psych. Ass'n 19 (Aug. 25, 2022), https://www.apa.org/news/apa/2022/social-media-children-teens. 29 Betul Keles et al., A systematic review: the influence of social media on depression, anxiety and 20 psychological distress in adolescents, Int'l J. Adolescence & Youth (202) 25:1, 79–93 (Mar. 3, 2019), 21 https://www.researchgate.net/publication/331947590_A_systematic_review_the_influence_of_so cial_media_on_depression_anxiety_and_psychological_distress_in_adolescents. 22 30 Zara Abrams, Why young brains are especially vulnerable to social media, Am. Psych. Ass'n 23 (Aug. 25, 2022), https://www.apa.org/news/apa/2022/social-media-children-teens. 31 Nino Gugushvili et al., Facebook use intensity and depressive symptoms: a moderated 24 mediation model of problematic Facebook use, age, neuroticism, and extraversion at 3, BMC Psych. 10, 279 (2022), https://doi.org/10.1186/s40359-022-00990-7. 25 32 Nino Gugushvili et al., Facebook use intensity and depressive symptoms: a moderated mediation model of problematic Facebook use, age, neuroticism, and extraversion at 3, BMC 26 Psych. 10, 279 (2022), https://doi.org/10.1186/s40359-022-00990-7. 33 27 Heiner Boettger, & Deborah Koeltezsch, , The fear factor: Xenoglossophobia or how to overcome the anxiety of speaking foreign languages, 4, Training Language and Culture, 43-55 28 (June 2020), https://www.researchgate.net/figure/Development-of-the-cortex-functions-The- PFC_fig1_342501707. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 15 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 22 of 278 1 2 3 4 5 6 7 8 9 10 67. Because of the immaturity of their prefrontal cortex, children have less impulse 11 control, and less ability to regulate their responses to social rewards, than adults. 12 68. Beginning around the age of 10, the brain also begins to change in important ways. 13 Specifically, the receptors for dopamine multiply in the subcortical region of the brain. 34 Dopamine 14 is a neurotransmitter that is central to the brain's reward system. 35 15 69. During this developmental phase, the brain learns to seek out stimuli (e.g., 16 Instagram) that result in a reward (e.g., likes) and cause dopamine to flood the brain's reward 17 pathways. Each time this happens, associations between the stimulus and the reward become 18 stronger.36 Notably, once the brain has learned to make this association, dopaminergic neurons 19 "shift their … activation from the time of reward delivery to the time of presentation of [a] 20 predictive cue."37 In other words, the anticipation of a reward can itself trigger a dopamine rush. 21 70. As New York University professor and social psychologist Adam Alter has 22 explained, product features such as "Likes" give users a dopamine hit similar to drugs and alcohol: 23 34 24 Zara Abrams, Why young brains are especially vulnerable to social media, Am. Psych. Ass'n (Aug. 25, 2022), https://www.apa.org/news/apa/2022/social-media-children-teens. 25 35 Zara Abrams, Why young brains are especially vulnerable to social media, Am. Psych. Ass'n (Aug. 25, 2022), https://www.apa.org/news/apa/2022/social-media-children-teens. 26 36 See Bryo Adinoff, Neurobiologic processes in drug reward and addiction, 12(6) Harv. Rev. Psychiatry 305-320 (2004), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1920543/. 27 37 Luisa Speranza et al., Dopamine: The Neuromodulator of Long-Term Synaptic Plasticity, 28 Reward and Movement Control, 10 Cells 735 (2021), https://pubmed.ncbi.nlm.nih.gov/33810328/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 16 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 23 of 278 1 "The minute you take a drug, drink alcohol, smoke a cigarette . . . when you get a like on social 2 media, all of those experiences produce dopamine, which is a chemical that's associated with 3 pleasure. When someone likes an Instagram post, or any content that you share, it's a little bit like 4 taking a drug. As far as your brain is concerned, it's a very similar experience." 38 5 71. When the release of dopamine in young brains is manipulated by Defendants' 6 products, it interferes with the brain's development and can have long-term impacts on an 7 individual's memory, affective processing, reasoning, planning, attention, inhibitory control, and 8 risk-reward calibration. 9 72. Given their limited capacity to self-regulate and their vulnerability to peer pressure, 10 children (including teens) are at greater risk of developing a mental disorder from use of 11 Defendants' products.39 Children are more susceptible than adults to feelings of withdrawal when 12 a dopamine hit wears off. Depending on the intensity, delivery, and timing of the stimulus, and the 13 severity of its withdrawal, these feelings can include anxiety, dysphoria, and irritability. 40 Children 14 also are more likely to engage in compulsive behaviors to avoid these symptoms, due to their 15 limited capacity for self-regulation, relative lack of impulse control, and struggle to delay 16 gratification. 17 73. This can lead to a vicious cycle. Repeated spikes of dopamine over time may cause 18 a child to build up a tolerance for the stimulus. In this process of "neuroadaptation," the production 19 20 21 38 Eames Yates, What happens to your brain when you get a like on Instagram, Business Insider 22 (Mar. 25, 2017), https://www.businessinsider.com/what-happens-to-your-brain-like-instagram- dopamine-2017-3; see also Sӧren Krach et al., The rewarding nature of social interactions, 4(22) 23 Frontiers in Behav. Neuro., (28 May 2010), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2889690/pdf/fnbeh-04-00022.pdf; Julian 24 Morgans, The Secret Ways Social Media Is Built for Addiction, Vice (May 17, 2017), https://www.vice.com/en/article/vv5jkb/the-secret-ways-social-media-is-built-for-addiction. 25 39 Betul Keles et al., A systematic review: the influence of social media on depression, anxiety and psychological distress in adolescents, Int'l J. Adolescence & Youth (202) 25:1, 79–93 (Mar. 3, 26 2019),https://www.researchgate.net/publication/331947590_A_systematic_review_the_influence _of_social_media_on_depression_anxiety_and_psychological_distress_in_adolescents 27 40 George Koob, and Nora Volkow. Neurobiology of addiction: a neurocircuitry analysis, 3 (8) 28 Lancet Psychiatry 760-773 (2016), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6135092/pdf/nihms-985499.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 17 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 24 of 278 1 of dopamine and the sensitivity of dopamine receptors are both reduced. 41 As a consequence, the 2 child requires more and more of the stimulus to feel the same reward. Worse, this cycle can cause 3 decreases in activity in the prefrontal cortex, leading to further impairments of decision-making 4 and executive functioning.42 5 74. As described further below, each Defendant deliberately designed, engineered, and 6 implemented dangerous features in their apps that present social-reward and other stimuli in a 7 manner that has caused Plaintiffs and many scores of others to compulsively seek out those stimuli, 8 develop negative symptoms when they were withdrawn, and exhibit reduced impulse control and 9 emotional regulation. 10 75. In short, children find it particularly difficult to exercise the self-control required to 11 regulate their use of Defendants' platforms, given the stimuli and rewards embedded in those apps, 12 and as a foreseeable consequence tend to engage in addictive and compulsive use. 43 13 3. Defendants designed their apps to attract and addict youth. 14 76. Instagram, Facebook, TikTok, Snap, and YouTube employ many similar defective 15 and dangerous product features that are engineered to induce more use by young people—creating 16 an unreasonable risk of compulsive use and addiction.44 For instance, all five apps harvest user data 17 and use this information to generate and push algorithmically tailored "feeds" of photos and videos. 18 And all five include methods through which approval can be expressed and received, such as likes, 19 20 41 21 George Koob, and Nora Volkow. Neurobiology of addiction: a neurocircuitry analysis, 3 (8) Lancet Psychiatry 760-773 (2016), 22 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6135092/pdf/nihms-985499.pdf. 42 George Koob & Nora Volkow. Neurobiology of addiction: a neurocircuitry analysis, 3 (8) 23 Lancet Psychiatry 760-773 (2016), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6135092/pdf/nihms-985499.pdf. 24 43 Fulton Crews et al., Adolescent cortical development: a critical period of vulnerability for addiction, 86 Pharmacology, Biochemistry and Behavior 189-199 (2007), 25 https://www.sciencedirect.com/science/article/pii/S009130570600400X. 44 26 See Kevin Hurler, For Sites Like Instagram and Twitter, Imitation Is the Only Form of Flattery, Gizmodo (Aug. 16, 2022), https://gizmodo.com/instagram-tiktok-snapchat-facebook-meta- 27 1849395419 ("Over the last decade, some of the most popular social media apps have blatantly ripped off features from some of the other most popular social media apps, in a tech version of 28 Capture the Flag where the only losers are the users who are forced to persist through this cat- and-mouse game."). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 18 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 25 of 278 1 hearts, comments, shares, or reposts. This section explains the psychological and social 2 mechanisms exploited by these and other product defects. 3 77. First, Defendants' apps are designed and engineered to methodically, but 4 unpredictably, space out dopamine-triggering rewards with dopamine gaps. The unpredictability is 5 key because, paradoxically, intermittent variable rewards (or "IVR") create stronger associations 6 (conditioned changes in the neural pathway) than fixed rewards. Products that use this technique 7 are highly addictive or habit forming. 8 78. IVR is based on insights from behavioral science dating back to research in the 9 1950s by Harvard psychologist B. F. Skinner. Skinner found that laboratory mice respond most 10 voraciously to unpredictable rewards. In one famous experiment, mice that pushed a lever received 11 a variable reward (a small treat, a large treat, or no treat at all). Compared with mice who received 12 the same treat every time, the mice who received only occasional rewards were more likely to 13 exhibit addictive behaviors such as pressing the lever compulsively. This exploitation of neural 14 circuitry is exactly how addictive products like slot machines keep users coming back. 15 79. The IVR aspect of slot machines is limited by the fact that they deliver rewards in a 16 randomized manner, irrespective of the person pulling the lever. By contrast, Defendants' apps are 17 designed to purposely withhold and release rewards on a schedule its algorithms have determined 18 is optimal to heighten a specific user's craving and keep them using the product. For example, 19 TikTok will at times delay a video it knows a user will like until the moment before it anticipates 20 the user would otherwise log out. Instagram's notification algorithm will at times determine that a 21 particular user's engagement will be maximized if the app withholds "Likes" on their posts and 22 then later delivers them in a large burst of notifications. 23 80. Defendants' use of IVR is particularly effective on and dangerous for adolescents, 24 given the incomplete aspects of their brain maturation described above—including lack of impulse 25 control and reduced executive functions. 26 81. There are multiple types of dopamine neurons that are connected with distinct brain 27 networks and have distinct roles in motivational control. Apart from the dopamine reward loop 28 triggered by positive feedback, other dopamine neurons are impacted by salient but non-rewarding PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 19 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 26 of 278 1 stimuli and even painful-aversive stimuli.45 Defendants' apps capitalize on this by algorithmically 2 ranking photos and videos that "engage" users because they present a dopamine pay-off, including 3 novel, aversive, and alarming images. 4 82. Second, there are multiple types of dopamine neurons that are connected with 5 distinct brain networks and have distinct roles in motivational control. Apart from the dopamine 6 reward loop triggered by positive feedback, other dopamine neurons are impacted by salient but 7 non-rewarding stimuli and even painful-aversive stimuli.46 Defendants' apps capitalize on this by 8 algorithmically ranking photos and videos that "engage" users because they present a dopamine 9 pay-off, including novel, aversive, and alarming images. 10 83. Third, dangerous and defective features in Defendants' apps manipulate young users 11 through their exploitation of "reciprocity"—the psychological phenomenon by which people 12 respond to positive or hostile actions in kind. Reciprocity means that people respond in a friendly 13 manner to friendly actions, and with negative retaliation to hostile actions. 47 Phillip Kunz best 14 illustrated the powerful effect of reciprocity through an experiment using holiday cards. Kunz sent 15 cards to a group of complete strangers, including pictures of his family and a brief note. 48 People 16 whom he had never met or communicated with before reciprocated, flooding him with holiday 17 cards in return.49 Most of the responses did not even ask Mr. Kunz who he was—they simply 18 responded to his initial gesture with a reciprocal action.50 19 84. Products like Instagram and Snapchat exploit reciprocity by, for example, 20 automatically telling a sender when their message is seen, instead of letting the recipient avoid 21 45 J.P.H. Verharen, Yichen Zhu, and Stephan Lammel Aversion hot spots in the dopamine system 64 Neurobiology 46-52 (March 5, 2020) https://doi.org/10.1016/j.conb.2020.02.002. 22 46 J.P.H. Verharen, Yichen Zhu, and Stephan Lammel Aversion hot spots in the dopamine system 23 64 Neurobiology 46-52 (March 5, 2020) https://doi.org/10.1016/j.conb.2020.02.002. 47 Ernst Fehr & Simon Gächter, Fairness and Retaliation: The Economics of Reciprocity, 14(3) J. 24 Econ. Persps. 159–81 (2000), https://www.researchgate.net/profile/Ernst-Fehr- 2/publication/23756527_Fairness_and_Retaliation_The_Economics_of_Reciprocity/links/5eb024 25 e945851592d6b87d3b/Fairness-and-Retaliation-The-Economics-of-Reciprocity.pdf. 48 Phillip R. Kunz & Michael Woolcott, Season's Greetings: From my status to yours, 5(3) Soc. 26 Sci. Rsch. 269–78 (Sept. 1976), https://doi.org/10.1016/0049-089X(76)90003-X. 49 Phillip R. Kunz & Michael Woolcott, Season's Greetings: From my status to yours, 5(3) Soc. 27 Sci. Rsch. 269–78 (Sept. 1976), https://doi.org/10.1016/0049-089X(76)90003-X. 50 28 Phillip R. Kunz & Michael Woolcott, Season's Greetings: From my status to yours, 5(3) Soc. Sci. Rsch. 269–78 (Sept. 1976), https://doi.org/10.1016/0049-089X(76)90003-X. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 20 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 27 of 278 1 disclosing whether it was viewed. Consequently, the recipient feels more obligated to respond 2 immediately, keeping users on the product.51 3 85. Fourth, Defendants' apps addict young users by preying on their already-heightened 4 need for social comparison and interpersonal feedback-seeking. 52 Because of their relatively 5 undeveloped prefrontal cortex, young people are already predisposed to status anxieties, beauty 6 comparisons, and a desire for social validation. 53 Defendants' apps encourage repetitive usage by 7 dramatically amplifying those insecurities. 8 86. Mitch Prinstein, Chief Science Officer for the American Psychology Association, 9 has explained that online and real-world interactions are fundamentally different. 54 For example, in 10 the real world, no public ledger tallies the number of consecutive days friends speak. Similarly, 11 "[a]fter you walk away from a regular conversation, you don't know if the other person liked it, or 12 if anyone else liked it."55 By contrast, a product defect like the "Snap Streak" creates exactly such 13 artificial forms of feedback.56 On Defendants' apps, friends and even complete strangers can deliver 14 (or withhold) dopamine-laced likes, comments, views, or follows. 57 15 87. The "Like" feature on Facebook, Instagram, TikTok, and YouTube, or other 16 comparable features common to Defendants' products, have an especially powerful effect on 17 51 Von Tristan Harris, The Slot Machine in Your Pocket, Spiegel Int'l (July 27, 2016), 18 https://www.spiegel.de/international/zeitgeist/smartphone-addiction-is-part-of-the-design-a- 1104237.html. 19 52 Jacqueline Nesi & Mitchell J Prinstein, Using Social Media for Social Comparison and Feedback-Seeking: Gender and Popularity Moderate Associations with Depressive Symptoms, 43 20 J. Abnormal Child Psych. 1427–38 (2015), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5985443/. 21 53 Susan Harter, The Construction of the Self: Developmental and Sociocultural Foundations 22 (Guilford Press, 2d ed., 2012) (explaining how, as adolescents move toward developing cohesive self-identities, they typically engage in greater levels of social comparison and interpersonal 23 feedback-seeking). 54 Zara Abrams, Why young brains are especially vulnerable to social media, Am. Psych. Ass'n 24 (Aug. 25, 2022), https://www.apa.org/news/apa/2022/social-media-children-teens. 55 Zara Abrams, Why young brains are especially vulnerable to social media, Am. Psych. Ass'n 25 (Aug. 25, 2022), https://www.apa.org/news/apa/2022/social-media-children-teens. 56 A "Snap Streak" is designed to measure a user's Snapchat activity with another user. Two users 26 achieve a "Snap Streak" when they exchange at least one Snap in three consecutive 24-hour periods. When successively longer "Streaks" are achieved, users are rewarded with varying tiers 27 of emojis. See infra p. 156. 57 28 Zara Abrams, Why young brains are especially vulnerable to social media, Am. Psych. Ass'n (Aug. 25, 2022), https://www.apa.org/news/apa/2022/social-media-children-teens. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 21 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 28 of 278 1 teenagers and can neurologically alter their perception of online posts. Researchers at UCLA used 2 magnetic resonance imaging to study the brains of teenage girls as they used Instagram. They found 3 that girls' perception of a photo changed depending on the number of likes it had generated. 58 That 4 an image was highly liked—regardless of its content—instinctively caused the girls to prefer it. As 5 the researchers put it, teens react to perceived "endorsements," even if likes on social media are 6 often fake, purchased, or manufactured. 59 7 88. The design of Defendants' apps also encourages unhealthy, negative social 8 comparisons, which in turn cause body image issues and related mental and physical disorders. 9 Given adolescents' naturally vacillating levels of self-esteem, they are already predisposed to 10 comparing "upward" to celebrities, influencers, and peers they perceive as more popular. 60 11 Defendants' apps turbocharge this phenomenon. On Defendants' apps, users disproportionately 12 post "idealized" content,61 misrepresenting their lives. That is made worse by appearance-altering 13 filters built into Defendants' apps, which underscore conventional (and often racially biased) 14 standards of beauty, by allowing users to remove blemishes, make bodies and faces appear thinner, 15 and lighten skin-tone. Defendants' apps provide a continuous stream of these filtered and fake 16 58 Lauren E. Sherman et al., The Power of the Like in Adolescence: Effects of Peer Influence on 17 Neural and Behavioral Responses to Social Media, 27(7) Psychol Sci. 1027 (2016),https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5387999/ 18 59 Lauren E. Sherman et al., The Power of the Like in Adolescence: Effects of Peer Influence on Neural and Behavioral Responses to Social Media, 27(7) Psychol Sci. 1027 (2016), 19 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5387999/; see also Stuart Wolpert, The teenage brain on social media, UCLA Newsroom (May 31, 2016), 20 https://newsroom.ucla.edu/releases/the-teenage-brain-on-social-media. 60 Jacqueline Nesi & Mitchell J Prinstein, Using Social Media for Social Comparison and 21 Feedback-Seeking: Gender and Popularity Moderate Associations with Depressive Symptoms, 43 J. Abnormal Child Psych. 1427–38 (2015), 22 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5985443/ ("Upward comparison occurs when people compare themselves to someone they perceive to be superior[ ], whereas a downward 23 comparison is defined by making a comparison with someone perceived to be inferior[.]"); Jin- Liang wang, Hai-Zhen Wang, James Gaskin, & Skyler Hawk, The Mediating Roles of Upward 24 Social Comparison and Self-esteem and the Moderating Role of Social Comparison Orientation in the Association between Social Networking Site Usage and Subjective Well-Being, Frontiers in 25 Psychology (May 2017), https://www.frontiersin.org/articles/10.3389/fpsyg.2017.00771/full#:~:text=Social%20compariso 26 n%20can%20be%20upward,inferior%20(Wills%2C%201981). 61 27 Jacqueline Nesi & Mitchell J Prinstein, Using Social Media for Social Comparison and Feedback-Seeking: Gender and Popularity Moderate Associations with Depressive Symptoms, 43 28 J. Abnormal Child Psych. 1427–38 (2015), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5985443/; PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 22 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 29 of 278 1 appearances and experiences.62 That encourages harmful body image comparisons by adolescents, 2 who begin to negatively perceive their own appearance and believe their bodies, and indeed their 3 lives, are comparatively worse.63 4 89. Fifth, Defendants' respective product features work in combination to create and 5 maintain a user's "flow-state": a hyper-focused, hypnotic state, where bodily movements are 6 reflexive and the user is totally immersed in smoothly rotating through aspects of the social media 7 product.64 8 90. As discussed in more detail below, defective features like the ones just described 9 can cause or contribute to (and, with respect to Plaintiffs, have caused and contributed to) the 10 following injuries in young people: eating and feeding disorders; depressive disorders; anxiety 11 disorders; sleep disorders; trauma- and stressor-related disorders; obsessive-compulsive and related 12 disorders; disruptive, impulse-control, and conduct disorders; suicidal ideation; self-harm; and 13 suicide.65 14 15 16 17 62 18 Jin Kyun Lee, The Effects of Social Comparison Orientation on Psychological Well-Being in Social Networking Sites: Serial Mediation of Perceived Social Support and Self-Esteem, Current 19 Psychology (2020), https://link.springer.com/content/pdf/10.1007/s12144-020-01114-3.pdf. 63 . Jin Kyun Lee, The Effects of Social Comparison Orientation on Psychological Well-Being in 20 Social Networking Sites: Serial Mediation of Perceived Social Support and Self-Esteem, Current Psychology (2020), https://link.springer.com/content/pdf/10.1007/s12144-020-01114-3.pdf. See 21 also Nino Gugushvili et al., Facebook use intensity and depressive symptoms: a moderated mediation model of problematic Facebook use, age, neuroticism, and extraversion at 3, BMC 22 Psych. 10, 279 (2022), https://doi.org/10.1186/s40359-022-00990-7 (explaining that youth are particularly vulnerable because they "use social networking sites for construing their identity, 23 developing a sense of belonging, and for comparison with others"); Jin Lee, The effects of social comparison orientation on psychological well-being in social networking sites: serial mediation 24 of perceived social support and self-esteem, 41 Current Psychology 6247-6259 (2022), https://doi.org/10.1007/s12144-020-01114-3. 25 64 See e.g., What Makes TikTok so Addictive?: An Analysis of the Mechanisms Underlying the World's Latest Social Media Craze, Brown Undergraduate J. of Pub. Health 26 (2021), https://sites.brown.edu/publichealthjournal/2021/12/13/tiktok/(describing how IVR and infinite scrolling may induce a flow state in users). 27 65 E.g., Nino Gugushvili et al., Facebook use intensity and depressive symptoms: a moderated 28 mediation model of problematic Facebook use, age, neuroticism, and extraversion at 3, BMC Psych. 10, 279 (2022), https://doi.org/10.1186/s40359-022-00990-7 (collecting sources). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 23 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 30 of 278 1 4. Millions of kids use Defendants' products compulsively. 2 91. Defendants have been staggeringly successful in their efforts to attract young users 3 to their apps. In 2021, 32% of 7- to 9-year-olds,66 49% of 10- to 12-year-olds,67 and 90% of 13- to 4 17-year-olds in the United States used social media.68 A majority of U.S. teens use Instagram, 5 TikTok, Snapchat, and/or YouTube. Thirty-two percent say they "wouldn't want to live without" 6 YouTube, while 20% said the same about Snapchat, and 13% said the same about both TikTok and 7 Instagram.69 8 92. U.S. teenagers who use Defendants' products are likely to use them every day. 9 Sixty-two percent of U.S. children ages 13-18 use social media daily. 70 And daily use often means 10 constant use. About one-in-five U.S. teens visit or use YouTube "almost constantly," while about 11 one-in-six report comparable usage of Instagram.71 Nearly half of U.S. teens use TikTok at least 12 13 14 15 16 66 Sharing Too Soon? Children and Social Media Apps, C.S. Mott Child's Hosp. Univ. Mich. 17 Health (Oct. 18, 2021), https://mottpoll.org/sites/default/files/documents/101821_SocialMedia.pdf. 18 67 Sharing Too Soon? Children and Social Media Apps, C.S. Mott Child's Hosp. Univ. Mich. 19 Health (Oct. 18, 2021), https://mottpoll.org/sites/default/files/documents/101821_SocialMedia.pdf. 68 20 Social Media and Teens, Am. Acad. Child & Adolescent Psychiatry (Mar. 2018), https://www.aacap.org/AACAP/Families_and_Youth/Facts_for_Families/FFF-Guide/Social- 21 Media-and-Teens-100.aspx; see also Victoria Rideout et al., The Common Sense Census: Media Use by Tweens and Teens, 2021 at 5, Common Sense Media (2022), 22 https://www.commonsensemedia.org/sites/default/files/research/report/8-18-census-integrated- report-final-web_0.pdf. 23 69 Victoria Rideout et al., Common Sense Census: Media use by tweens and teens, 2021 at 31, Common Sense Media (2022), 24 https://www.commonsensemedia.org/sites/default/files/research/report/8-18-census-integrated- report-final-web_0.pdf. 25 70 Victoria Rideout et al., Common Sense Census: Media use by tweens and teens, 2021 at 31, Common Sense Media (2022), 26 https://www.commonsensemedia.org/sites/default/files/research/report/8-18-census-integrated- report-final-web_0.pdf. 27 71 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 28 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 2022/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 24 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 31 of 278 1 "several times a day."72 In one study, U.S. teenage users reported checking Snapchat thirty times a 2 day on average.73 3 93. Teenagers know they are addicted to Defendants' products: 36% admit they spend 4 too much time on social media.74 Yet they can't stop. Of the teens who use at least one social media 5 product "almost constantly," 71% say quitting would be hard. Nearly one-third of this population— 6 and nearly one-in-five of all teens—say quitting would be "very hard."75 7 94. Notably, the more teens use Defendants' apps, the harder it is to quit. Teens who 8 say they spend too much time on social media are almost twice as likely to say that giving up social 9 media would be hard, compared to teens who see their social media usage as about right. 76 10 95. Despite using social media frequently, most young people don't particularly enjoy 11 it. In 2021, only 27% of boys and 42% of girls ages 8-18 reported liking social media "a lot." 77 12 Moreover, one survey found that young people think social media is the main reason youth mental 13 health is getting worse.78 About twice as many of the surveyed youth believed that social media is 14 15 16 72 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 17 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 2022/. 18 73 Erinn Murphy et al., Taking Stock with Teens, Fall 2021 at 13, Piper Sandler (2021), https://tinyurl.com/89ct4p88; see also Emily Vogels et al., Teens, Social Media and Technology 19 2022, Pew Rsch. Ctr. (Aug. 10, 2022), https://www.pewresearch.org/internet/2022/08/10/teens- social-media-and-technology-2022/. 20 74 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 21 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 2022/. 75 22 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 23 2022/. 76 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 24 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 2022/. 25 77 Victoria Rideout et al., Common Sense Census: Media use by tweens and teens, 2021 at 34, Common Sense Media (2022), 26 https://www.commonsensemedia.org/sites/default/files/research/report/8-18-census-integrated- report-final-web_0.pdf. 27 78 Headspace (2018) National youth mental health survey 2018, National Youth Mental Health 28 Foundation (2018), https://headspace.org.au/assets/headspace-National-Youth-Mental-Health- Survey-2018.pdf PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 25 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 32 of 278 1 the main reason for declining mental health than the next likely cause, and over seven times more 2 believed it to be the main cause rather than drugs and alcohol. 79 3 5. Defendants' apps have created a youth mental health crisis. 4 96. Nearly a decade of scientific and medical studies demonstrate that dangerous 5 features engineered into Defendants' platforms—particularly when used multiple hours a day— 6 can have a "detrimental effect on the psychological health of [their] users," including compulsive 7 use, addiction, body dissatisfaction, anxiety, depression, and self-harming behaviors such as eating 8 disorders.80 9 97. Addiction and compulsive use of Defendants' products can entail a variety of 10 behavioral problems including but not limited to: (1) a lessening of control, (2) persistent, 11 compulsive seeking out of access to the product, (3) using the product more, and for longer, than 12 intended, (4) trying to cut down on use but being unable to do so, (5) experiencing intense cravings 13 or urges to use, (6) tolerance (needing more of the product to achieve the same desired effect), (7) 14 developing withdrawal symptoms when not using the product, or when the product is taken away, 15 (8) neglecting responsibilities at home, work, or school because of the intensity of usage, (9) 16 continuing to use the product even when doing so interferes and causes problems with important 17 family and social relationships, (10) giving up important or desirable social and recreational 18 activities due to use, and (11) continuing to use despite the product causing significant harm to the 19 user's physical and mental health. 20 98. Many of these injuries can be long-lasting, if not lifelong. For example, the long- 21 term effects of eating disorders can include: (1) dermatological effects to the nails and hair; 22 (2) gastrointestinal illnesses, such as gastroparesis or hypomotility of the colon; (3) impacts to the 23 endocrine system, such as glycolic or metabolic conditions, bone loss, and hormonal conditions; 24 79 Headspace National Youth Mental Health Survey 2018, National Youth Mental Health Foundation (2018), https://headspace.org.au/assets/headspace-National-Youth-Mental-Health- 25 Survey-2018.pdf (surveying more than 4,000 Australians ages 12-25). 80 26 See, e.g., Fazida Karim et al., Social Media Use and Its Connection to Mental Health: A Systemic Review, Cureus Volume 12(6) (June 15, 2020), 27 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7364393/; Alexandra R. Lonergan et al., Protect me from my selfie: Examining the association between photo-based social media behaviors and 28 self-reported eating disorders in adolescence, Int. J. of Eating Disorders 756 (Apr. 7, 2020), https://onlinelibrary.wiley.com/doi/epdf/10.1002/eat.23256. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 26 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 33 of 278 1 (4) nervous system effects, such as gray matter brain loss or atrophy; (5) skeletal system effects, 2 such as bone loss; (6) cardiovascular effects, such as structural heart damage, mitral valve prolapse, 3 or fluid around the heart; and (7) fertility issues.81 4 99. Each Defendant has long been aware of this research, but chose to ignore or brush 5 it off.82 For example, in 2018, Meta employees mocked it as "BS . . . pseudo science," [sic] and "a 6 bunch of people trying to get air time."83 Yet, as discussed at length below, Defendants conducted 7 some of the research themselves—and then hid their unfavorable findings from the public. 84 8 100. Scientists have studied the impacts of the overuse of social media since at least 2008, 9 with social media addiction recognized in literature around that time after a pervasive upsurge in 10 Facebook use.85 The Bergen Social Media Addiction Scale assesses social media addiction along 11 six core elements: 1) salience (preoccupation with the activity), 2) mood modification (the behavior 12 alters the emotional state), 3) tolerance (increasing activity is needed for the same mood-altering 13 effects), 4) withdrawal (physical or psychological discomfort when the behavior is discontinued), 14 15 16 81 See, e.g., Anorexia Nervosa, Cleveland Clinic 17 https://my.clevelandclinic.org/health/diseases/9794-anorexia-nervosa#outlook--prognosis; Bulimia Nervosa; Cleveland Clinic https://my.clevelandclinic.org/health/diseases/9795-bulimia- 18 nervosa#symptoms-and-causes. 82 In August 2019, a social psychologist, and leading expert on the effect that technology products 19 have on the mental health of their users, wrote to Mr. Zuckerberg ahead of a meeting to note that a new study "point[ed] heavily to a connection, not just from correlational studies but from true 20 experiments, which strongly indicate[d] causation, not just correlation" between Meta's products and harms to users' wellbeing. META3047MDL-003-00089107 at META3047MDL-003- 21 000891078. In some cases, Meta was not only aware of research connecting its products to detrimental effects but actively sought to undermine it. See META3047MDL-003-00082165 at 22 META3047MDL-003-00082165 (discussing methods to undermine research on addiction to apps). 23 83 META3047MDL-003-00082165. 84 See, e.g., Haugen_00016373 at Haugen_00016381 ("The best external research indicates that 24 Facebook's impact on people's well-being is negative."); Haugen_00016373 at Haugen_00016414 (Mar. 9, 2020 presentation stating "All problematic users were experiencing 25 multiple life impacts," including loss of productivity, sleep disruption, relationship impacts, and safety risks); Haugen_00005458 at Haugen_00005500 (Sept. 18, 2019 presentation containing a 26 slide stating "But, We Make Body Image Issues Worse for 1 in 3 Teen Girls"). 85 27 Tim Davies & Pete Cranston, Youth Work and Social Networking: Interim Report, The National Youth Agency (May 2008), 28 https://www.researchgate.net/publication/233911484_Youth_Work_and_Social_Networking_Fin al_Research_Report. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 27 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 34 of 278 1 5) conflict (ceasing other activities or social interaction to perform the behavior), and 6) relapse 2 (resuming the behavior after attempting to control or discontinue it). 86 3 101. Beginning in at least 2014, researchers began demonstrating that addictive and 4 compulsive use of Defendants' apps leads to negative mental and physical outcomes for kids. 5 102. In 2014, a study of 10- to 12-year-old girls found that increased use of Facebook 6 was linked with body image concerns, the idealization of thinness, and increased dieting. 87 (This 7 study was sent to Mark Zuckerberg in 2018, in a letter signed by 118 public health advocates.) 88 8 103. In 2016, a study demonstrated that young people who frequently use Defendants' 9 apps are more likely to suffer sleep disturbances than their peers who use them infrequently. 89 10 Defendants' products, driven by IVR algorithms, deprive users of sleep by sending push 11 notifications and emails at night, prompting children to re-engage with the apps when they should 12 be sleeping. Disturbed and insufficient sleep is associated with poor health outcomes, 90 including 13 increased risk of major depression—by a factor of more than three— 91and future suicidal behavior 14 in adolescents.92 The American Academy of Sleep Medicine has recommended that, in a 24-hour 15 16 86 Cecilie Andreassen et al., The relationship between addictive use of social media and video 17 games and symptoms of psychiatric disorders: a large-scale cross-sectional study, 30(2) Psychol. of Addictive Behav., 252-262 (2016), http://dx.doi.org/10.1037/adb0000160. 18 87 Marika Tiggemann & Amy Slater, NetTweens: The Internet and body image concerns in preteenage girls, 34(5), J. Early Adolesc. 606-620 (June 2014), 19 https://journals.sagepub.com/doi/epub/10.1177/0272431613501083. 88 20 Campaign for a Commercial-Free Childhood, Letter to Mark Zuckerberg Re: Facebook Messenger Kids (Jan. 30, 2018), https://fairplayforkids.org/wp-content/uploads/archive/devel- 21 generate/gaw/FBMessengerKids.pdf. 89 Jessica C. Levenson et al., The Association Between Social Media Use and Sleep Disturbance 22 Among Young Adults, 85 Preventive Med. 36–41 (Apr. 2016), https://www.sciencedirect.com/science/article/abs/pii/S0091743516000025. 23 90 Jessica C. Levenson et al., The Association Between Social Media Use and Sleep Disturbance Among Young Adults, 85 Preventive Med. 36–41 (Apr. 2016), 24 https://www.sciencedirect.com/science/article/abs/pii/S0091743516000025; National Institute of Mental Health. 2016. The teen brain: 6 things to know, available at 25 https://www.nimh.nih.gov/health/publications/the-teen-brain-still-under-construction/index.shtml; R. Sather& A. Shelat, Understanding the teen brain, 26 https://www.urmc.rochester.edu/encyclopedia/content.aspx?ContentTypeID=1&ContentID=3051. 91 E. Roberts & H Doung, The Prospective Association between Sleep Deprivation and 27 Depression among Adolescents Sleep, Volume 37, Issue 2, 1 Feb. 2014. 92 28 X. Liu, D. Buysse, Sleep and youth suicidal behavior: a neglected field, Current Opinion in Psychiatry (May 2006). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 28 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 35 of 278 1 period, children aged 6–12 years should regularly sleep 9–12 hours and teenagers aged 13–18 years 2 should sleep 8–10 hours.93 3 104. In another 2016 study, 52% of girls said they use image filters every day, and 80% 4 reported using an app to change their appearance before the age of 13. 94 In fact, 77% of girls 5 reported trying to change or hide at least one part of their body before posting a photo of themselves, 6 and 50% believe they did not look good enough without editing. 95 7 105. In 2017, British researchers asked 1,500 teens to rate how Instagram, Snapchat, and 8 YouTube affected them on certain well-being measures, including anxiety, loneliness, body image, 9 and sleep.96 Teens rated all three platforms as having a negative impact on body image, "FOMO" 10 (fear of missing out), and sleep. Teens also noted that Instagram and Snapchat had a negative impact 11 on anxiety, depression, and loneliness. 12 106. In 2018, a Journal of Social and Clinical Psychology study examined a group of 13 college students whose use of Instagram, Facebook, and Snapchat was limited to 10 minutes per 14 day per platform. The study found that this limited-use group showed "significant reductions in 15 loneliness and depression over three weeks" compared to a control group that used social media as 16 usual.97 17 18 93 19 S. Paruthi, L. Brooks, C. D'Ambrosio, et al, Consensus statement of the American Academy of Sleep Medicine on the recommended amount of sleep for healthy children: methodology and 20 discussion, 12 J Clin Sleep Med 1549–61 (2016). 94 Anna Haines, From "Instagram Face" to "Snapchat Dysmorphia": How Beauty Filters Are 21 Changing the Way We See Ourselves, Forbes (Apr. 27, 2021), https://www.forbes.com/sites/annahaines/2021/04/27/from-instagram-face-to-snapchat- 22 dysmorphia-how-beauty-filters-are-changing-the-way-we-see-ourselves/?sh=3c32eb144eff. 95 Anna Haines, From "Instagram Face" to "Snapchat Dysmorphia": How Beauty Filters Are 23 Changing the Way We See Ourselves, Forbes (Apr. 27, 2021), https://www.forbes.com/sites/annahaines/2021/04/27/from-instagram-face-to-snapchat- 24 dysmorphia-how-beauty-filters-are-changing-the-way-we-see-ourselves/?sh=3c32eb144eff. 96 Royal Society for Public Health, #StatusOfMind, 25 https://www.rsph.org.uk/static/uploaded/d125b27c-0b62-41c5-a2c0155a8887cd01.pdf; see also Jonathan Haidt, The Dangerous Experiment on Teen Girls, The Atlantic (Nov. 21, 2021), 26 available at https://www.theatlantic.com/ideas/archive/2021/11/facebooks-dangerous- experiment-teen-girls/620767/. 27 97 Melissa G. Hunt et al., No More FOMO: Limiting Social Media Decreases Loneliness and 28 Depression, 37 J. of Social & Clinical Psych. (Dec. 5, 2018), https://guilfordjournals.com/doi/epdf/10.1521/jscp.2018.37.10.751. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 29 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 36 of 278 1 107. In 2018, a systematic literature review of nine studies published in the Indian 2 Journal of Psychiatry concluded that dangerous features in social networking platforms "contribute 3 to increased exposure to and engagement in self-harm behavior, as users tend to emulate self- 4 injurious behavior of others online, adopt self-injurious practices from self-harm videos, or are 5 encouraged and acclaimed by others, thus normalizing self-injurious thoughts and behavior." 98 6 108. A 2019 survey of American adolescents ages 12-14 found that a user's displeasure 7 with their body could be predicted based on their frequency of using social media (including 8 Instagram and Facebook) and based on the extent to which they engaged in behaviors that adopt an 9 observer's point-of-view (such as taking selfies or asking others to "rate one's looks"). This effect 10 was more pronounced among girls than boys.99 11 109. A third study in 2019 of more than 6500 American adolescents ranging in age from 12 12 to 15 years old found that those who used social media for 3 hours or more per day were more 13 likely to suffer from mental health problems such as anxiety and depression. 100 Notably, this 14 association remained significant even after adjusting for demographics, past alcohol and marijuana 15 use, and history of mental health problems.101 16 110. In 2020, a study of Australian adolescents found that investment in others' selfies 17 (through likes and comments) was associated with greater odds of meeting criteria for 18 clinical/subclinical bulimia nervosa, clinical/subclinical binge-eating disorder, night eating 19 syndrome, and unspecified feeding and eating disorders. 102 20 98 Aksha Memon et al., The role of online social networking on deliberate self-harm and 21 suicidality in adolescents: a systematized review of literature, 60(4) Indian J Psychiatry 384-392 (Oct-Dec 2018), http://10.4103/psychiatry.IndianJPsychiatry_414_17. 22 99 Ilyssa Salomon & Christia Spears Brown, The Selfie Generation: Examining the Relationship 23 Between Social Media Use and Early Adolescent Body Image, Journal of Early Adolescence (Apr. 21, 2018), https://journals.sagepub.com/doi/abs/10.1177/0272431618770809. 100 24 Kira Riehm et al., Associations between time spent using social media and internalizing and externalizing problems among US youth, 76(12) JAMA Psychiatry (2019), 25 https://jamanetwork.com/journals/jamapsychiatry/fullarticle/2749480. 101 Kira Riehm et al., Associations between time spent using social media and internalizing and 26 externalizing problems among US youth, 76(12) JAMA Psychiatry (2019), https://jamanetwork.com/journals/jamapsychiatry/fullarticle/2749480. 27 102 Alexandra R. Lonergan et al., Protect Me from My Selfie: Examining the Association Between 28 Photo-Based Social Media Behaviors and Self-Reported Eating Disorders in Adolescence, Int'l J. of Eating Disorders (Apr. 7, 2020), https://onlinelibrary.wiley.com/doi/epdf/10.1002/eat.23256. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 30 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 37 of 278 1 111. In 2020, a longitudinal study investigated whether "Facebook Addiction Disorder" 2 predicted suicide-related outcomes, and found that children and adolescents addicted to Facebook 3 are more likely to engage in self-injurious behavior, such as cutting and suicide. 103 4 112. In 2020, clinical research demonstrated an observable link between youth social 5 media use and disordered eating behavior. 104 The more time young girls spend using Defendants' 6 products, the more likely they are to develop disordered eating behaviors. 105 And the more social 7 media accounts adolescents have, the more disordered eating behaviors they exhibit. 106 8 9 10 11 12 13 14 15 16 17 18 19 20 103 21 See, e.g., Julia Brailovskaia et al., Positive mental health mediates the relationship between Facebook addiction disorder and suicide-related outcomes: a longitudinal approach, 00(00) 22 Cyberpsychology, Behavior, and Social Networking (2020), https://doi.org/10.1089/cyber.2019.0563; Jean M. Twenge et al., Increases in Depressive 23 Symptoms, Suicide-Related Outcomes, and Suicide Rates Among U.S. Adolescents After 2010 and Links to Increased New Media Screen Time, 6 Clinical Psych. Sci. 3–17 (2018). 24 104 Simon M. Wilksch et al., The relationship between social media use and disordered eating in young adolescents, 53 Int'l J. Eating Disorders 96–106 (2020), 25 https://pubmed.ncbi.nlm.nih.gov/31797420/. 105 Simon M. Wilksch et al., The relationship between social media use and disordered eating in 26 young adolescents, 53 Int'l J. Eating Disorders 96–106 (2020), https://pubmed.ncbi.nlm.nih.gov/31797420/. 27 106 Simon M. Wilksch et al., The relationship between social media use and disordered eating in 28 young adolescents, 53 Int'l J. Eating Disorders 96–106 (2020), https://pubmed.ncbi.nlm.nih.gov/31797420/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 31 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 38 of 278 1 113. Eating disorders often occur simultaneously with other self-harm behaviors such as 2 cutting and are often associated with suicide. 107 3 114. In a 2021 study, female undergraduates were randomly shown thinspiration (low 4 body mass index and not muscular), fitspiration (muscular and exercising), or neutral photos. 108 5 Thinspiration and fitspiration images lowered self-esteem, even in those with a self-perceived 6 healthy weight.109 7 115. A 2022 study of Italian adolescent girls (12-17) and young women (18-28) found 8 that Instagram's image editing and browsing features, combined with an emphasis on influencer 9 interactions, promulgated unattainable body ideals that caused users to compare their bodies to 10 those ideals.110 These trends were more prominent among adolescent girls, given their higher 11 susceptibility to social pressures related to their bodies and given the physical changes associated 12 with puberty. 13 116. In 2023, a study of magnetic resonance images demonstrated that compulsive use 14 of Defendants' apps measurably alters children's brains.111 This study measured fMRI responses 15 in 12-year-old adolescents who used Facebook, Instagram, and Snapchat over a three-year period 16 and found that neural patterns diverged. Specifically, those who engaged in high social media 17 checking behavior "showed lower neural sensitivity to social anticipation" than those who engaged 18 in low to moderate checking behavior.112 19 107 Sonja Swanson et al., Prevalence and correlates of eating disorders in adolescents, 68(7) Arch Gen Psychiatry 717-723 (2011), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5546800/. 20 108 Karikarn Chansiri & Thipkanok Wongphothiphan, The indirect effects of Instagram images on 21 women's self-esteem: The moderating roles of BMI and perceived weight, 00(0) New Media & Society 1-23 (2021), https://journals.sagepub.com/doi/epub/10.1177/14614448211029975. 109 22 Karikarn Chansiri & Thipkanok Wongphothiphan, The indirect effects of Instagram images on women's self-esteem: The moderating roles of BMI and perceived weight, 00(0) New Media & 23 Society 1-23 (2021), https://journals.sagepub.com/doi/epub/10.1177/14614448211029975. 110 Federica Pedalino and Anne-Linda Camerini, Instagram use and body dissatisfaction: The 24 mediating role of upward social comparison with peers and influencers among young females, 19(3) Int'l J of Environmental Research and Public Health 1543 (2022), 25 https://www.mdpi.com/1660-4601/19/3/1543. 111 Maria Maza et al., Association of habitual checking behaviors on social media with 26 longitudinal functional brain development, JAMA Ped., (Jan. 3, 2023), https://jamanetwork.com/journals/jamapediatrics/article-abstract/2799812. 27 112 Maria Maza et al., Association of habitual checking behaviors on social media with 28 longitudinal functional brain development, JAMA Ped., (Jan. 3, 2023), https://jamanetwork.com/journals/jamapediatrics/article-abstract/2799812. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 32 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 39 of 278 1 117. Defendants' apps have triggered depression, anxiety, eating disorders, self-harm, 2 and suicidality among thousands of children, including the Plaintiffs in this action. Defendants have 3 created nothing short of a national crisis. 4 118. From 2009 to 2019, the rate of high school students who reported persistent sadness 5 or hopelessness increased by 40% (to one out of every three kids).113 The share of kids who 6 seriously considered suicide increased by 36%, and those that created a suicide plan increased by 7 44%.114 8 119. From 2007 to 2019, suicide rates among youth aged 10-24 in the United States 9 increased by 57%.115 10 120. From 2007 to 2016, emergency room visits for youth aged 5-17 rose 117% for 11 anxiety disorders, 44% for mood disorders, and 40% for attention disorders. 116 12 121. By 2019, one-in-five children aged 3-17 in the United States had a mental, 13 emotional, developmental, or behavioral disorder. 117 Mental health issues are particularly acute 14 among females.118 15 16 113 Protecting Youth Mental Health: The U.S. Surgeon General's Advisory at 8, U.S. Dep't Health 17 & Hum. Servs. (Dec. 7, 2021), https://www.hhs.gov/sites/default/files/surgeon-general-youth- mental-health-advisory.pdf. 18 114 Protecting Youth Mental Health: The U.S. Surgeon General's Advisory at 8, U.S. Dep't Health 19 & Hum. Servs. (Dec. 7, 2021), https://www.hhs.gov/sites/default/files/surgeon-general-youth- mental-health-advisory.pdf. 115 20 Protecting Youth Mental Health: The U.S. Surgeon General's Advisory at 8, U.S. Dep't Health & Hum. Servs. (Dec. 7, 2021), https://www.hhs.gov/sites/default/files/surgeon-general-youth- 21 mental-health-advisory.pdf. 116 Charmaine Lo, Children's mental health emergency department visits: 2007-2016, 145(6) 22 Pediatrics e20191536 (June 2020), https://doi.org/10.1542/peds.2019-1536. 117 U.S. Surgeon General Issues Advisory on Youth Mental Health Crisis Further Exposed by 23 COVID-19 Pandemic, U.S. Dep't Health & Hum. Servs. (Dec. 14, 2021), https://adasoutheast.org/u-s-surgeon-general-issues-advisory-on-youth-mental-health-crisis- 24 further-exposed-by-covid-19-pandemic/; see also Jean M. Twenge et al., Increases in Depressive Symptoms, Suicide-Related Outcomes, and Suicide Rates Among U.S. Adolescents After 2010 and 25 Links to Increased New Media Screen Time, 6 Clinical Psych. Sci. 3–17 (2017), https://doi.org/10.1177/216770261772337 (noting that mental health issues are particularly acute 26 among females). 118 27 U.S. Surgeon General Issues Advisory on Youth Mental Health Crisis Further Exposed by COVID-19 Pandemic, U.S. Dep't Health & Hum. Servs. (Dec. 14, 2021), 28 https://adasoutheast.org/u-s-surgeon-general-issues-advisory-on-youth-mental-health-crisis- further-exposed-by-covid-19-pandemic/ PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 33 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 40 of 278 1 122. On December 7, 2021, the United States Surgeon General issued an advisory on the 2 youth mental health crisis.119 The Surgeon General explained, "[m]ental health challenges in 3 children, adolescents, and young adults are real and widespread. Even before the pandemic, an 4 alarming number of young people struggled with feelings of helplessness, depression, and thoughts 5 of suicide—and rates have increased over the past decade." 120 Those "mental health challenges 6 were the leading cause of disability and poor life outcomes in young people." 121 7 123. On February 13, 2023, the CDC released new statistics revealing that, in 2021, one 8 in three girls seriously considered attempting suicide.122 9 124. As discussed herein, each of the Defendants' products manipulates minor users' 10 brains by building in stimuli and social reward mechanisms (e.g., "Likes") that cause users, such 11 as Plaintiffs, to compulsively seek social rewards. That, in turn, leads to neuroadaptation; a child 12 requires more and more stimuli to obtain the desired dopamine release, along with further 13 impairments of decision-making. It also leads to reward-seeking through increasingly extreme 14 content, which is more likely to generate intense reactions from other users. These consequences 15 are the foreseeable results of Defendants' engineering decisions. 16 6. Defendants' defective products encourage dangerous "challenges." 17 125. Dangerous "viral" challenges are one particularly pernicious result of the defective 18 design of Defendants' apps. "Online challenges or dares typically involve people recording 19 themselves doing something difficult, which they share online to encourage others to repeat." 123 20 119 U.S. Surgeon General Issues Advisory on Youth Mental Health Crisis Further Exposed by 21 COVID-19 Pandemic, U.S. Dep't Health & Hum. Servs. (Dec. 14, 2021), https://adasoutheast.org/u-s-surgeon-general-issues-advisory-on-youth-mental-health-crisis- 22 further-exposed-by-covid-19-pandemic/. 120 U.S. Surgeon General Issues Advisory on Youth Mental Health Crisis Further Exposed by 23 COVID-19 Pandemic, U.S. Dep't Health & Hum. Servs. (Dec. 14, 2021), https://adasoutheast.org/u-s-surgeon-general-issues-advisory-on-youth-mental-health-crisis- 24 further-exposed-by-covid-19-pandemic/ 121 U.S. Surgeon General Issues Advisory on Youth Mental Health Crisis Further Exposed by 25 COVID-19 Pandemic, U.S. Dep't Health & Hum. Servs. (Dec. 14, 2021), https://adasoutheast.org/u-s-surgeon-general-issues-advisory-on-youth-mental-health-crisis- 26 further-exposed-by-covid-19-pandemic/ 122 Azeen Ghorayashi & Roni Caryn Rabin, Teen Girls Report Record Levels of Sadness, C.D.C. 27 Finds, N.Y. Times (Feb. 13, 2023),https://www.nytimes.com/2023/02/13/health/teen-girls- sadness-suicide-violence.html. 28 123 TikTok, Online Challenges, https://www.tiktok.com/safety/en-us/online-challenges/.. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 34 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 41 of 278 1 These challenges often generate significant engagement—sometimes millions of likes or views— 2 and resulting social rewards to the users who post videos of themselves carrying out the challenges. 3 ByteDance product research has found that the number one most identified reason for teen 4 participation in challenges is "[g]etting views/likes/comments," followed by "[i]mpressing others 5 online."124 Predictably, a substantial portion of online challenges—created for the purpose of 6 generating social rewards—are very dangerous. 7 126. For example, one common social media challenge is the "Blackout Challenge," 8 where youth are encouraged to make themselves faint by holding their breath and constricting their 9 chest muscles, or by restricting airflow with a ligature around their neck. This challenge is 10 dangerous because, should the participant fail to remove the ligature around their neck prior to 11 fainting, they may strangle themselves. Similarly, an "I Killed Myself" challenge involves 12 participants faking their own deaths to record their family members' reactions upon believing their 13 loved one has died. This challenge is dangerous because certain methods of participating can 14 actually kill (or inflict catastrophic injury) on participants. Likewise, the game Russian Roulette— 15 in which a participant loads a revolver with a single bullet, spins the chamber until it falls on a 16 random slot, and then shoots themselves—has taken on new life as a social media challenge. 17 Certain plaintiffs in the MDL have participated in each of these challenges and suffered death or 18 catastrophic injury as a result. 19 127. Again, these injuries and deaths are a foreseeable consequence of Defendants' 20 addictive product designs. Many other addictive products cause injury or death because 21 neuroadaptation causes addicts to use increasingly extreme methods to maintain dopamine levels. 22 That compulsive use of social media would do the same was, at all relevant times, foreseeable, 23 particularly as to young users whose abilities to assess risks, make wise decisions, and regulate 24 their responses to perceived social needs are still developing. 25 128. Defendants are perfectly aware of the foreseeable risks to youth presented by their 26 apps' "viral" promotion of dangerous challenges. As ByteDance internal documents observe, 27 124 TIKTOK3047MDL-001-00002975. Similarly, ByteDance internal documents have identified 28 "a desire for competition, status, followers, clicks and to push limits" as being reasons young people participate in dangerous challenges. TikTok3047MDL-001-00002375. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 35 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 42 of 278 1 "[y]oung people are more susceptible to copying a dangerous challenge than adults because, 2 developmentally, their ability to weigh up risk and to think beyond the immediate consequences of 3 their actions is not yet fully formed. They are also more likely to take risks, to make impulsive 4 choices, and to overestimate their ability to cope with risk. . . . The ability to understand the finality 5 of death is also not fully fledged." 125 6 129. Defendants have encouraged the viral challenge phenomenon in spite of the fact that 7 their encouragement furthers dangerous challenges themselves and a broader ecosystem in which 8 those dangerous challenges occur. 9 130. Meta, TikTok, and YouTube use engagement-optimized algorithms to control users' 10 main feeds. Such algorithms spread extreme content as a consequence of its propensity to generate 11 engagement. That design defect foreseeably leads to dangerous challenges spreading easily on these 12 Defendants' platforms. In addition, Meta's algorithms were at one point in time structured to triple 13 amplification of posts that used the word "challenge" in a hashtag—which Meta product engineers 14 would later observe was "super high risk" and prompted "#killchallenge" to go viral. 126 15 131. Defendants have further encouraged challenges in other ways. ByteDance regularly 16 creates overlays and filters that facilitate viral challenges. It offers advertisers the ability to launch 17 Branded Hashtag Challenges and promotes them on user feeds. 127 It boasts that challenges are 18 "geared towards building awareness and engagement," and that "research shows that they can 19 deliver strong results"—i.e., increased return on ad spending—"at every stage of the funnel." This, 20 in turn, generates advertising revenue for ByteDance. ByteDance employees have described 21 challenges as "one of our core features." 128 22 132. Snap also promotes viral challenges through Snapchat's Snap Spotlight feature. It 23 gives cash prizes to challenge participants whose challenges receive the most views on Snap 24 25 125 TikTok3047MDL-001-00002375. 26 126 META3047MDL-003-00030070. 127 TikTok for Business, Branded Hashtag Challenge: Harness the Power of Participation (Mar. 27 16, 2022), https://www.tiktok.com/business/en-US/blog/branded-hashtag-challenge-harness-the- power-of-participation. 28 128 TikTok3047MDL-001-00000177. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 36 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 43 of 278 1 Spotlight.129 It has also created overlays that encourage challenges—such as a Speed Filter, 2 showing how fast a given user was going at the time they took a particular Snap. 130 Other 3 Defendants have also promoted viral challenges based on a recognition that such challenges drive 4 engagement and advertising revenue. 5 7. Defendants' defective social media apps facilitate and contribute to the sexual exploitation and sextortion of children, and the ongoing 6 production and spread of child sex abuse material online. 7 133. It is well documented that sexual predators use Defendants' products to target and 8 exploit minors. They are drawn to social media because it provides them with easy access to a large 9 pool of potential victims, many of whom are addicted to Defendants' products. On February 7, 10 2023, for example, the FBI and international law enforcement agencies issued a joint warning about 11 the global "financial sextortion crisis" which stated: "Financial sextortion can happen anywhere, 12 although it mainly occurs on the digital platforms where children are already spending their screen 13 time, like social media and gaming websites, or video chat applications. On these platforms, 14 predators often pose as girls of a similar age and use fake accounts to target young boys, deceiving 15 them into sending explicit photos or videos. The predator then threatens to release the 16 compromising materials unless the victim sends payment, however, in many cases, the predator 17 will release the images anyway."131 18 134. Rather than mitigate the risk of sexual exploitation and harm to children and teens 19 on their products, Defendants have facilitated and exacerbated it by implementing defective product 20 features that help sexual predators connect with children. Defendants' products are designed in 21 unsafe ways—including, as to some or all Defendants, flawed age verification, lack of meaningful 22 mechanisms to prevent sham accounts, default-public profiles, matching and recommending 23 connections between adults and minors, promoting unsolicited messages and interactions from 24 25 129 "Snapchat adds 'challenges' with cash prizes to its TikTok competitor," The Verge, Oct. 6, 2021, https://www.theverge.com/2021/10/6/22711632/snapchat-spotlight-challenges-announced- 26 cash-prizes-tiktok. 130 See Lemmon v. Snap, Inc., 995 F.3d 1085, 1085 (9th Cir. 2021). 27 131 International Law Enforcement Agencies Issue Joint Warning about global financial sextortion 28 crisis, FBI (2023), https://www.fbi.gov/news/press-releases/international-law-enforcement- agencies-issue-joint-warning-about-global-financial-sextortion-crisis. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 37 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 44 of 278 1 adults, and wholly inadequate and ineffective parental controls, among others—that allow children 2 to be easily identified, targeted, accessed, and exploited. 3 135. Compounding the problem, Defendants routinely fail to report abuse. Steve Grocki, 4 the U.S. Department of Justice's Chief of the Criminal Division's Child Exploitation & Obscenity 5 Section, put it this way: "A lot of times, someone has come across something problematic and the 6 platform isn't doing anything[.]" He went on to say, "[t]hese reports can be of great value because 7 they signal where there are big problems and we can flag those issues to Internet companies, such 8 as when products are being exploited by offenders, they aren't meeting reporting requirements, or 9 when children under the age restriction are accessing inappropriate content." 132 10 136. By failing to implement adequate age and identity verification on the products, 11 Defendants knowingly and foreseeably place children in the pathways of sexual predators, who 12 utilize their products and exploit their defective design features. Age verification on Defendants' 13 products can be defective for a variety of reasons, including: 14 Inaccurate information: Users can easily enter false information, such as a 15 fake date of birth, to bypass age restrictions; 16 No measures to prevent sham accounts: Defendants' products are structured 17 so that users can easily create multiple accounts under different names and 18 ages; and 19 Incomplete implementation: Defendants' products only partially implement 20 age verification, such as requiring users to provide their date of birth but not 21 verifying it through any means. 22 137. To be sure, Defendants possess the technology to identify minors who are posing as 23 adults and adults who are posing as minors, but they do not use this information to identify violative 24 accounts and remove them from their products. 25 138. For example, by making minors' profiles public by default, certain Defendants have 26 supplied sexual predators with detailed background information about children, including their 27 132 28 Equality Now, Steve Grocki Expert Interview - United States (Nov. 15, 2021), https://www.equalitynow.org/stories/steve-grocki-expert-interview-united-states/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 38 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 45 of 278 1 friends, activities, interests, and even location. By providing this information, these Defendants put 2 children at risk of being approached and befriended by sexual predators. These young targets had 3 no idea that their posts, friends, pictures, and general online sharing represented a windfall of 4 information that a predator could use to sexually exploit or abuse them. 5 139. Meta, ByteDance, and Google are aware of these very real risks that public-by- 6 default accounts for minors can cause by leaving minors broadly exposed to adult strangers, and 7 yet have only recently (in response to Plaintiffs' actions) begun to create default privacy settings 8 for youth accounts—even though they have been aware of these harmful interactions for years. 9 140. Defendants have also lagged behind in restricting who can reach minor users 10 through Direct Message features. That is particularly egregious in the case of Meta, whose own 11 research has made clear that the vast majority of unwanted interactions received by these accounts 12 come from unconnected adults.133 13 141. Defendants' parental controls are also defective in giving parents effective control 14 over their children's online activity, which can lead to kids connecting with predators. These defects 15 take several forms: 16 Limited scope: parental control tools only partially cover a child's activity, 17 leaving gaps that predators can exploit; 18 Inadequate monitoring: parental control tools do not provide real-time 19 monitoring of a child's activity, meaning that harmful interactions with 20 predators go unnoticed; 21 Lack of customization: parents are not able to fully customize their parental 22 control settings to meet the specific needs of their family and children, 23 leaving them with a one-size-fits-all solution that is wholly ineffective in 24 preventing connections with predators; and 25 Opt-in format: parental control tools that require parents to affirmatively link 26 to their child's account are futile when parents are not notified that their 27 minor child has opened an account on the platform in the first place. 28 133 META3047MDL-003-00107197. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 39 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 46 of 278 1 142. Defendants are well aware that vulnerable young users—whom Defendants induce 2 to spend large amounts of time on their products, through a powerful combination of algorithmic 3 recommendations and addictive features designed to make it hard for a user to disengage—are 4 uniquely susceptible to grooming by seasoned sexual predators. Research shows that young users 5 are heavily reliant on their social connections—exploring and shaping their identity through their 6 social relationships. 7 143. Defendants' defective product features have benefited sexual predators in many 8 other ways as well. For example, sexual predators leverage Defendants' use of ephemeral, 9 "disappearing" technology to assure young users that there is no risk to them sending a sexual photo 10 or video. Trusting young users are then horrified to discover that these videos have been captured 11 by predators and then circulated to their own friends and contacts or other sexual predators. In some 12 severe cases, young users find themselves in the nightmarish scheme known as "sextortion," where 13 a predator threatens to circulate the sexual images of the minor unless the predator is paid to keep 14 the images under wraps. Law enforcement agencies across the country report that this scheme has 15 become pervasive on Defendants' products. 16 144. As a direct and foreseeable consequence of Defendants' connecting children to 17 sexual predators, Defendants' products facilitate and increase the risk of sexual exploitation, sexual 18 abuse, and sextortion of children. 134 19 145. Many adult predators use Defendants' apps to facilitate commercial sex acts, 20 including the production and distribution of known Child Sex Abuse Material ("CSAM"). 135 21 22 134 23 "Sextortion is a form of sexual exploitation. The offender uses coercion and threats to compel the victim to produce sexual images or videos engaging in sexual acts. Sometimes the offender 24 already possesses nude or sexual images of the victim, and is threatening to release them if the victim will not do as the offender commands." United States Attorney's Office Southern District 25 of Indiana, What is Sextortion?, https://www.justice.gov/media/844516/dl?inline. 135 26 The term CSAM includes any definition of "Child Pornography" under 18 U.S.C. § 2256(8) and corresponding caselaw, including but not limited to, United States v. Dost, 636 F. Supp. 828, 27 832 (S.D. Cal. 1986), aff'd sub nom. United States v. Wiegand, 812 F.2d 1239 (9th Cir. 1987), and aff'd, 813 F.2d 1231 (9th Cir. 1987); see also Canadian Centre for Child Protection Report, 28 Project Arachind: Online Availability of Chis Sexual Abuse Material, https://protectchildren.ca/pdfs/C3P_ProjectArachnidReport_Summary_en.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 40 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 47 of 278 1 CSAM is illegal contraband and constitutes a personal injury. 136 Defendants bear responsibility for 2 the continued creation and distribution of CSAM on their apps, which they know or should know 3 about.137 "The unlawful conduct of everyone who reproduces, distributes, or possesses the images 4 of the victim's abuse … plays a part in sustaining and aggravating this tragedy." 138 5 146. Tragically, for the young users who have been groomed and cajoled into creating or 6 sending CSAM, these images are rapidly shared, reproduced, and disseminated to other sexual 7 predators at an astonishing rate. A 2021 article published in Fast Company revealed, "Facebook 8 noted that more than 90% of the reported CSAM content on its platforms was the 'same as or visibly 9 similar to previously reported content,' which is the crux of the problem. Once a piece of CSAM 10 content is uploaded, it spreads like wildfire…." 139 Each time their images are disseminated it 11 represents a fresh revictimization of young users who are unable to stop the proliferation and spread 12 of their images. Despite this knowledge, Defendants failed to design their apps with readily 13 available technology or less exploitative features that in conjunction with each other could curb, 14 rather than fuel, the sexual exploitation and sextortion of adolescent users, as well as limit the 15 distribution of known CSAM on their products. 16 147. Tools like PhotoDNA and YouTube's CSAI Match are image comparison 17 technologies that can detect matches between modified versions of the same image or images. 18 Consider two versions of the same image: one in full color, the other in black and white. To the 19 human eye, these are understood to be images depicting the same thing, but to a computer, they are 20 entirely different. These tools return values indicating the "closeness" between two images. 140 This 21 process, sometimes called "fuzzy matching" or "perceptual hashing," looks at the visual content of 22 the image instead of the exact binary image data (i.e., the digital fingerprint or cryptographic hash). 23 136 24 See Amy v. Curtis, 2020 WL 5365979 (N.D. Cal. Sept. 8, 2020) (extending In re Boland, 946 F.3d 335 (6th Cir. 2020)). 137 25 New York v. Ferber, 458 U.S. 747, 759 (1982). 138 Paroline v. United States, 572 U.S. 434, 457 (2014). 26 139 Glen Pounder & Rasty Turek, On social media, child sexual abuse material spreads faster than it can be taken down, Fast Company (July 14, 2021), 27 https://www.fastcompany.com/90654692/on-social-media-child-sexual-abuse-material-spreads- faster-than-it-can-be-taken-down. 28 140 Microsoft, How does PhotoDNA work?, https://www.microsoft.com/en-us/photodna. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 41 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 48 of 278 1 148. Although Defendants have used PhotoDNA and other hash-matching technology to 2 compare suspect images against databases of previously verified CSAM provided by law 3 enforcement or child protection agencies, 141 their product design modifications have diminished 4 the effectiveness and success of these hash-matching technologies and detection models. Hash- 5 matching detection can be made more effective by, for example, prioritizing factors like "tolerance" 6 (i.e., specificity) over "efficiency" (i.e., cost and scalability) and "distinctness" (i.e., more 7 specificity).142 8 149. In addition, Defendants have defectively designed their products' reporting tools so 9 that they actually impede the ability of young users and their parents to report CSAM or material 10 related to child exploitation. A 2020 report by the Canadian Centre for Child Protection revealed 11 that social media companies design their products in a way that makes reporting CSAM difficult, 12 if not impossible, for users.143 While CSAM can be reported as generally harmful material, 13 Defendants designed their products with no specific function to report CSAM contraband. 14 Defendants could easily do so. 15 150. Plaintiffs' reports of CSAM, and requests to remove and take down such material 16 from Defendants' products, often fall on deaf ears. In many cases, Defendants have failed to take 17 action until legal correspondence was sent. 144 18 141 19 This is a process called "hashing." For a fuller explanation, see Hany Farid, An Overview of Perceptual Hashing, J. of Online Trust and Safety (Oct. 2021) at 2-3, 20 https://tsjournal.org/index.php/jots/article/download/24/14 and Jennifer Langston, How photodna for video is being used to fight online child exploitation On the Issues (2020), 21 https://news.microsoft.com/on-the-issues/2018/09/12/how-photodna-for-video-is-being-used-to- fight-online-child-exploitation/. 22 142 Hany Farid, An Overview of Perceptual Hashing, Journal of Online Trust and Safety (October 2021) at 2-3, https://tsjournal.org/index.php/jots/article/download/24/14 ("In practice, the choice 23 of a hash is based on a number of factors, including: 1. Scale. When operating at the scale of a major social media platform, for example, with billions of daily uploads, any hash must be highly 24 efficient and distinct. At this scale, even a 1/100 or even 1/10,000 false positive rate (incorrectly matching two images) is untenable. 2. Tolerance. When trying to limit the upload of, for example, 25 legal adult pornography, resilience may be less important than, for example, trying to limit child sexual abuse imagery."). 26 143 Canadian Centre for Child Protection, Reviewing Child Sexual Abuse Material Reporting Functions on Popular Platforms, 27 https://protectchildren.ca/pdfs/C3P_ReviewingCSAMMaterialReporting_en.pdf. 144 28 Angus Crawford, Facebook failed to remove sexualised images of children, BBC News (Mar. 7, 2017), https://www.bbc.com/news/technology-39187929 . PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 42 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 49 of 278 1 151. For minor victims addicted to Defendants' products, the inability to report illegal 2 images as contraband delays their removal and results in repeated re-victimization. Experience 3 shows that minors become obsessed with monitoring the spread of their own CSAM in a desperate 4 attempt to control what often becomes an endless invasion of their privacy. 145 The Canadian Centre 5 for Child Protection reported that one CSAM victim stated, "I spend hours every day searching for 6 my own content, reporting thousands of accounts and posts sharing CSAM. When platforms don't 7 actively look for or prevent this content from being uploaded, the burden falls on me to have these 8 images removed."146 9 152. This design defect also hinders Defendants' ability to provide timely information to 10 child welfare and law enforcement agencies, even when Defendants know or should know children 11 are in immediate danger.147 12 153. Further, Defendants have unreasonably increased their products' dangerousness to 13 children by failing to utilize freely available and industry-proven child protection tools such as 14 Project Arachnid Shield, a CSAM detection platform that processes thousands of images per second 15 and sends removal notices to electronic services providers. Project Arachnid Shield is a free 16 Application Programming Interface ("API") that Defendants could incorporate into the design of 17 their apps, which would increase their safety and prevent child exploitation and the distribution of 18 CSAM through their products.148 Designing their products with such existing and free technology 19 would reduce the amount of CSAM being uploaded onto their products to be used as grooming 20 tools on Defendants' adolescent users. Again, to be clear, CSAM is contraband material. 21 154. In 2021, the Thorn organization published a report detailing the myriad problems of 22 child sexual exploitation that were present throughout Facebook, Instagram, TikTok, Snapchat and 23 24 145 Angus Crawford, Facebook failed to remove sexualised images of children, BBC News (Mar. 25 7, 2017), https://www.bbc.com/news/technology-39187929. 146 Angus Crawford, Facebook failed to remove sexualised images of children, BBC News (Mar. 26 7, 2017), https://www.bbc.com/news/technology-39187929. 147 Canadian Centre for Child Protection, Resources & Research: Reviewing Child Sexual Abuse 27 Material Reporting Functions On Popular Platforms, https://protectchildren.ca/en/resources- research/csam-reporting-platforms/ (last visited Feb 9, 2023). 28 148 Project Arachnid, https://www.projectarachnid.ca/en/#shield. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 43 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 50 of 278 1 YouTube.149 Thorn contacted Meta and provided a topline read-out of its findings and offered to 2 meet with Meta to give the company a deeper look at the Thorn's data. 150 3 155. Thorn's key findings about the prevalence of sexual exploitation on Defendants' 4 platforms are startling: 5 "[O]ne-quarter of participants (25%) reported having had an online sexual 6 interaction with someone they thought was 18 or older;" 7 "Younger (9-13) children and LGBTQ+ children are at higher risk (9-13 year 8 olds 20% sexual experience with adult, 32% of LGBTQ youth);" 9 "Neither blocking nor reporting [offenders] protects minors from continued 10 harassment;" 11 "[R]ecidivism is a problem: over half of participants who had blocked (55%) 12 or reported (55%) someone, said they were recontacted online. Younger 13 boys are particularly at risk;" 14 "Kids reporting having difficulties finding the right reporting categories 15 when they did decide to report- they also noted that things like "involves a 16 child" doesn't resonate with them as teens;" 17 "17% of respondents who used FB say they have had a harmful experience, 18 10% said they had a sexual interaction on the platform;" 19 "26% of respondents who used IG say they have had a harmful experience, 20 16% said they have had a sexual interaction on the platform;" 21 "18% of respondents who used Messenger say they have had a harmful 22 experience, 11% said they had a sexual interaction on the platform. In 6% of 23 these sexual interactions the minor believed the person was 18%." 151 24 25 26 149 META3047MDL-003-00186841 (May 2021, Responding to Online Threats: Minors' Perspectives on Disclosing, Reporting and Blocking. Findings from 2020 Quantitative Research 27 among 9-17 year olds). 150 META3047MDL-003-00186838. 28 151 META3047MDL-003-00186838. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 44 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 51 of 278 1 8. Defendants could have avoided harming Plaintiffs. 2 156. Each Defendant solicited customers, including Plaintiffs, on the open market and 3 encouraged the use of their defective apps. 4 157. Each Defendant offers its app to the consuming public with dangerous, standardized 5 features and designs (discussed below) that users, like Plaintiffs, cannot bargain to change. 6 158. Plaintiffs (along with millions of other U.S. users) confer a benefit on each 7 Defendant in exchange for using their respective products. 8 159. Each Defendant could have, but purposefully failed to, design its products to protect 9 and avoid injury to kids and adolescent users, such as Plaintiffs. 10 160. Each Defendant knew or should have known that adolescents' developing brains 11 leave them relatively less able to delay gratification, control impulses, or resist immediately 12 pleasurable social rewards. 13 161. Each Defendant knew or should have known that the more children use social media, 14 the harder it is to quit. 15 162. Each Defendant knew or should have known that excessive use of its apps has severe 16 and wide-ranging effects on youth mental and physical health. 17 163. Each Defendant knew or should have known that youth are especially vulnerable to 18 long-term harm from its addictive products. 19 164. Each Defendant knew or should have known that the design of its products attracts, 20 enables, and facilitates child predators, and that such predators use its apps to recruit and sexually 21 exploit children for the production of CSAM and its distribution on Defendants' products. 22 165. Each Defendant knew or should have known that the longer adolescent users remain 23 engaged with its products, the higher the risk that adult predators will target them. 24 166. Each Defendant knew or should have known that many of its users are under the age 25 of 13, despite the limitations set out in COPPA. 26 167. Despite all that, each Defendant failed to adequately warn Plaintiffs and Consortium 27 Plaintiffs of the known risks and harms of using its products. Each Defendant avoided design 28 changes that would have increased youth safety. And each Defendant pressed ahead with changes PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 45 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 52 of 278 1 designed to keep kids hooked, even though they knew or should have known those changes were 2 risky. 3 168. Each Defendant was in a superior position to control the risks of harm, ensure the 4 safety of its apps, insure against the defects, and spread the costs of any harm resulting from the 5 defects. 6 169. Plaintiffs, Consortium Plaintiffs, and the consuming public did not have, and could 7 not have had, as much knowledge as Defendants about Defendants' apps and how they were 8 defectively designed. 9 170. Consumers, including Plaintiffs and Consortium Plaintiffs, could not have inspected 10 the apps before accepting them to learn of the defects or the harms that flow from the defects. 11 9. Defendants consistently refer to and treat their apps as products. 12 171. Each Defendant characterizes and treats their various apps as mass-produced, mass- 13 marketed products that each of the Defendants designs, tests, researches, builds, ships, markets, 14 and makes widely available in the stream of commerce for personal use by consumers, including 15 youth. 16 172. For example, Defendants routinely characterize their social media platforms as 17 products in their regulatory filings and communications with the financial markets and investors. 18 In its 2022 Annual Report, Meta stated that "[t]he term 'Family' refers to our Facebook, Instagram, 19 Messenger, and WhatsApp products," and that "there are inherent challenges in measuring usage 20 of our products across large online and mobile populations." 152 Similarly, in its 2015 Annual 21 Report, Google stated that its "core products such as … YouTube… each have over one billion 22 monthly active users."153 Likewise, in its 2022 Annual Report, Snap explains that its "flagship 23 product, Snapchat, is a visual messaging application." 154 24 152 Meta, 2022 Annual Report 3-4 (Feb. 2, 2023), https://investor.fb.com/financials/default.aspx. 25 153 Google, 2015 Annual Report 2 (Feb. 11, 2016), https://www.sec.gov/Archives/edgar/data/1288776/000165204416000012/goog10-k2015.htm. 26 154 Snap, Inc., Snap, Inc. 2022 Annual Report 10 (Jan. 31, 2023), 27 https://investor.snap.com/financials/Annual-Report/default.aspx; see also Snap Inc., Investor Letter Q3 2022 2 (Oct 20, 2022), 28 https://s25.q4cdn.com/442043304/files/doc_financials/2022/q3/Snap-Inc.-Q3-2022-Investor- PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 46 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 53 of 278 1 173. Defendants likewise routinely describe their apps as products in statements to public 2 officials and users. In testimony to the Senate Commerce and Judiciary Committees, Mark 3 Zuckerberg stated that Facebook's "controls are not just to make people feel safe; it's actually what 4 people want in the product."155 He noted that Facebook "want[s] our products to be valuable to 5 people."156 And he stated that, "fundamentally, at our core, [Meta is] a technology company where 6 the main thing that we do is have engineers and build products." 157 7 174. The other Defendants have made similar statements. In a written response to Senator 8 Marsha Blackburn, Snap noted that it takes suggestions into consideration "when releasing 9 products."158 In written testimony to the Senate Commerce Committee, a ByteDance witness 10 referred to the "variety of tools and controls we have built into the product." 159 YouTube executives 11 have used similar language. In written testimony to the Senate Commerce Committee, one 12 YouTube witness noted that consultants "work closely with the product teams to ensure that product 13 design reflects an understanding of children's unique needs." 160 And in written testimony to the 14 Senate Committee on Homeland Security and Governmental Affairs, YouTube's Chief Product 15 16 17 18 Letter-(10.20.2022).pdf, ("Our team remains focused on expanding product offering and deepening engagement with our global community"). 19 155 Bloomberg Government, Transcript of Mark Zuckerberg's Senate Hearing, Washington Post (Apr. 10, 2018), https://www.washingtonpost.com/news/the-switch/wp/2018/04/10/transcript-of- 20 mark-zuckerbergs-senate-hearing/. 156 21 Bloomberg Government, Transcript of Mark Zuckerberg's Senate Hearing, Washington Post (Apr. 10, 2018), https://www.washingtonpost.com/news/the-switch/wp/2018/04/10/transcript-of- 22 mark-zuckerbergs-senate-hearing/. 157 Bloomberg Government, Transcript of Mark Zuckerberg's Senate Hearing, Washington Post 23 (Apr. 10, 2018), https://www.washingtonpost.com/news/the-switch/wp/2018/04/10/transcript-of- mark-zuckerbergs-senate-hearing/. 24 158 SNAP0000246. 159 Protecting Kids Online: Snapchat, TikTok, and Youtube: Hearing Before the Subcomm. on 25 Consumer Protection, Product Safety, and Data Security of the S. Comm. On Commerce, Science, and Transportation (Oct. 26, 2022), https://www.commerce.senate.gov/2021/10/protecting-kids- 26 online-snapchat-tiktok-and-youtube. 160 27 Protecting Kids Online: Snapchat, TikTok, and Youtube: Hearing Before the Subcomm. on Consumer Protection, Product Safety, and Data Security of the S. Comm. On Commerce, Science, 28 and Transportation (Oct. 26, 2022), https://www.commerce.senate.gov/2021/10/protecting-kids- online-snapchat-tiktok-and-youtube. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 47 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 54 of 278 1 Officer stated that "responsibility is our top priority at YouTube and informs every product and 2 policy decision we make."161 3 175. Defendants employ "product managers" and have established "product teams" 4 responsible for the development, management, operation, and marketing of their apps. For example, 5 Meta lists on the careers section of its website a role for "Product Manager, Instagram Camera." 162 6 Snap's website lists job openings for a "Product Marketing Manager, App Ads" and a "Director of 7 Product Management, Ad Marketplace and Quality."163 TikTok Careers has employment 8 opportunities for a Livestream Product Manager," "Senior Product Manager-Operation Platform," 9 "Vertical Product Marketing Manager," and "Technical Product Specialist – Platforms." 164 Earlier 10 this year, YouTube Careers was hiring for a "Director Product Management, YouTube Shorts 11 Discovery."165 YouTube's pitch: "Make products as fun as they are useful." 166 12 176. Defendants understand that, when they are developing their apps, they are building, 13 testing, doing quality control on, and modifying their "products." For instance, in a 2013 earnings 14 call, one Meta employee noted, "We will continue to focus our development efforts to build 15 products that drive engagement for people of all ages." 167 In a 2012 interview at Tech Crunch 16 Disrupt, Zuckerberg noted that Instagram "has a super talented group of, of engineers. They're 17 building [this] amazing product." Meta employees often complement each other's "great product 18 improvement[s]."168 19 161 20 Social Media's Impact on Homeland Security, Part II: Hearing Before the S. Comm. On Homeland Security and Governmental Affairs (Sept. 14, 2022) (written testimony of Neal Mohan, 21 Chief Product Officer, YouTube and SVP, Google), https://www.hsgac.senate.gov/wp- content/uploads/imo/media/doc/Testimony-Mohan-2022-09-14.pdf. 22 162 Meta Careers, https://www.metacareers.com/. 163 Snap Inc., Jobs, https://snap.com/en-US/jobs. 23 164 TikTok, Careers, https://careers.tiktok.com/. 165 24 YouTube, Careers, https://www.youtube.com/jobs/; see also Kevin Roose, YouTube's Product Chief on Online Radicalization and Algorithmic Rabbit Holes, N.Y. Times (Mar. 29, 2019), 25 https://www.nytimes.com/2019/03/29/technology/youtube-online-extremism.html (interview with "YouTube's chief product officer": "our product teams here are thinking of all of these 26 solutions"). 166 YouTube, YouTube Jobs, https://www.youtube.com/jobs/product-and-design/. 27 167 Facebook, Facebook Q3 2013 Earnings Call, Zuckerberg Transcripts 236 (Oct. 20, 2013), https://epublications.marquette.edu/zuckerberg_files_transcripts/236. 28 168 Haugen_000020610. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 48 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 55 of 278 1 177. Other companies operate similarly. In a blog post, Snap referred to its "rebuild" of 2 the Snapchat "Android product."169 YouTube asked its users for "Your Help to Test New Product 3 Features," saying the "main goal of this study is to test new product features to better understand 4 your needs."170 Likewise, ByteDance's internal documents routinely refer to TikTok as a product, 5 including in references to plans for a "Product Feature Livestream," "[c]omplet[ing] the team 6 building of product [and] basic figures of product," the importance of "reviewing product 7 issues,"171 and various "Product operations" issues including "translat[ion] into English." 172 8 178. Defendants also set up workflows and systems that "package" and "ship" their apps 9 as "products." Zuckerberg has explained how "we ship a lot of tweaks to the products, or small 10 changes to existing products."173 ByteDance has referred to "special product packaging 11 awareness."174 12 179. Defendants treat their apps as products in customer research, branding, marketing, 13 and growth discussions. In text messages with Kevin Systrom in 2012, Zuckerberg noted, "I'm 14 really excited about what we can do to grow Instagram as an independent brand and product." 175 15 In an internal document from 2018, Instagram employees noted that only "[a] few more days will 16 be needed before we have an idea of how good our product-market fit is." 176 Similarly, TikTok's 17 Product Policies note that "[e]ach product has its own set of guidelines … but they are adjusted to 18 reflect specific product's mission and vision." 177 Google has been equally candid in referring to 19 20 169 Snap Newsroom, Restructuring and Refocusing our Business (Sept. 11, 2022) Restructuring and Refocusing our Business. 21 170 YouTube Official Blog, We Need Your Help to Test New Product Features, 22 https://blog.youtube/news-and-events/we-need-your-help-to-test-new-product/ (2012). 171 TikTok3047MDL-001-00000769. 23 172 TikTok3047MDL-001-00058090. 173 Facebook, Facebook Q3 2013 Earnings Call" (2013). Zuckerberg Transcripts at 2. 24 https://epublications.marquette.edu/zuckerberg_files_transcripts/236 174 25 TikTok3047MDL-001-00000215. 175 Facebook, Facebook text log between Mark Zuckerberg and Kevin Systrom, (2012) 26 https://epublications.marquette.edu/zuckerberg_files_transcripts?utm_source=epublications.marq uette.edu%2Fzuckerberg_files_transcripts%2F1330&utm_medium=PDF&utm_campaign=PDFC 27 overPages 176 META3047MDL-003-000031888. 28 177 TikTok3047MDL-001-00060877. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 49 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 56 of 278 1 YouTube as a "product," publishing an anniversary post entitled, "A Look Back as We Move 2 Forward: YouTube Product Launches in 2011."178 3 180. Meta has characterized Instagram and Facebook as "products" when discussing the 4 harms and injuries that those apps inflict on users. Meta described as a "product" issue the role of 5 Instagram's "Explore" feature in elevating the risk of suicide and self-injury in certain users. 179 6 Meta employees have characterized as a "product" issue users' addictive use of Instagram. 180 Still 7 another Meta employee has expressed concern about Facebook, stating "I'm anxious about whether 8 FB the product is good for the world." 181 9 B. FACTUAL ALLEGATIONS AS TO META 10 1. Background and overview of Meta's products. 11 181. Meta operates and designs Facebook and Instagram, two of the world's most popular 12 social media products. In 2022, two billion users worldwide were active on Instagram each month, 13 and almost three billion were monthly active users of Facebook. 182 This enormous reach has been 14 accompanied by enormous damage for Plaintiffs and other adolescent users. 15 182. Meta understands that its products are used by kids under 13: "there are definitely 16 kids this age on IG [Instagram] . . . they aren't supposed to be on IG at all." 183 It understands that 17 its products are addictive: "(1) teens feel addicted to IG and feel a pressure to be present, (2) like 18 addicts, they feel that they are unable to stop themselves from being on IG, and (3) the tools we 19 currently have aren't effective at limiting their time on the ap (sic)." 184 It understands that addictive 20 use leads to problems: "it just keeps people coming back even when it stops being good for 21 178 YouTube Official Blog, A Look Back as We Move Forward: YouTube Product Launches in 22 2011, https://blog.youtube/news-and-events/look-back-as-we-move-forward-youtube/ (Jan. 23, 2012). 23 179 META3047MDL-003-00068863 at META3047MDL-003-00068883. 180 24 Haugen_00010127 ("It seems clear from what's presented here that some of our users are addicted to our products. And I worry that driving sessions incentivizes us to make our products 25 more addictive."). 181 Haugen_00012484 at Haugen_ 00012553 26 182 Alex Barinka, Meta's Instagram Users Reach 2 Billion, Closing In on Facebook, Bloomberg (Oct. 26, 2022), https://www.bloomberg.com/news/articles/2022-10-26/meta-s-instagram-users- 27 reach-2-billion-closing-in-on-facebook. 183 META3047MDL-003-00123666 at META3047MDL-003-00123666. 28 184 META3047MDL-003-00157036 at META3047MDL-003-00157036. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 50 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 57 of 278 1 them."185 And it understands that these problems can be so extreme as to include encounters 2 between adults and minors—with such "sex-talk" 32x more prevalent on Instagram than on 3 Facebook186 and driven largely by a single product defect. 187 4 183. Despite this knowledge, Meta has abjectly failed at protecting child users of 5 Instagram and Facebook. Rather than stand up a five-alarm effort to stop the problems created by 6 its products, "the mental health team stopped doing things . . . it was defunded . . . completely 7 stopped."188 "We've consistently deprioritized addiction as a work area." 189 Zuckerberg himself 8 was warned personally: "We are not on track to succeed for our core well-being topics (problematic 9 use, bullying & harassment, connections, and SSI), and are at increased regulatory risk and external 10 criticism. These affect everyone, especially Youth and Creators; if not addressed, these will follow 11 us into the Metaverse. . . ."190 12 184. Yet Meta did nothing. And its reason was simple: "the growth impact was too 13 high."191 Taking action would lower usage of (and therefore lower profits earned from) a critical 14 audience segment. "Youth and Teens are critically important to Instagram . . . there's a new group 15 of 13-year-olds every year and the competition over their Social Media engagement has never been 16 more fierce."192 17 185. Meta's frequent gestures towards youth safety were never serious and always driven 18 by public relations: "it's all theatre."193 Meta offered tools to kids and parents, like "time spent," 19 that it knew presented false data—"Our data as currently shown is incorrect. . . . We're sharing bad 20 metrics externally. . . . we vouch for these numbers."194 At the same time, Meta engaged in a cynical 21 185 META3047MDL-003-00011760 at META3047MDL-003-00011761. 22 186 META3047MDL-003-00119838 at META3047MDL-003-00119838. 187 23 META3047MDL-003-00013254 at META3047MDL-003-00013255. This defect, "People You May Know," allows adults to connect with minors on Facebook and Instagram. 24 188 META3047MDL-003-00011697 at META3047MDL-003-00011698. 189 META3047MDL-003-00157133 at META3047MDL-003-00157134. 25 190 META3047MDL-003-00188109 at META3047MDL-003-00188114 (footnote omitted). 26 "SSI" refers to "suicide and self-injury." META3047MDL-003-00068863. 191 META3047MDL-003-00013254 at META3047MDL-003-00013254. 27 192 META3047MDL-003-00030070 at META3047MDL-003-00030071. 193 META3047MDL-003-00053803 at META3047MDL-003-00053803. 28 194 META3047MDL-003-00157133. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 51 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 58 of 278 1 campaign to "counter-messag[e] around the addiction narrative" by discrediting existing research 2 as "completely made up. . ."195 Meta knew better. Indeed, both Zuckerberg and Instagram CEO 3 Adam Mosseri heard firsthand from a leading researcher that Instagram and Facebook posed unique 4 dangers to young people. Meta could have but failed to prevent the harms suffered by Plaintiffs. 5 a. Meta's origins and the development of Facebook. 6 186. In October 2003, a sophomore at Harvard College named Mark Zuckerberg hacked 7 into the websites of Harvard's residential dorms to collect photos of students. He then designed a 8 website called "Facemash" that invited users to rank the "hotness" of female students by comparing 9 their photos side-by-side. In just one day, Facemash users cast over 22,000 votes judging the looks 10 of women at Harvard.196 This was precisely the point of Facemash, as its homepage made clear: 11 "Were we let in for our looks? No. Will we be judged on them? Yes." When interviewed about 12 Facemash, Zuckerberg stated, "I'm a programmer and I'm interested in the algorithms and math 13 behind it." Zuckerberg was summoned to appear before Harvard's disciplinary body. 14 187. After narrowly escaping expulsion, Zuckerberg began writing code for a new 15 website, thefacebook.com. The growth of the product that subsequently became Facebook has been 16 extensively documented and was the subject of an Academy Award-winning film. 197 By the end of 17 2005, Facebook had expanded its reach to thousands of colleges and high schools in the United 18 States and abroad. Over the coming years, Facebook grew well beyond campuses, reaching over 19 100 million total active users by the fall of 2008. By February 2011, Facebook had become the 20 largest online photo host, holding nearly 100 billion photos. 198 And by the end of 2011, Facebook, 21 195 META3047MDL-003-00082165 at META3047MDL-003-00082165 - META3047MDL-003- 22 00082166. 196 Katherine Kaplan, Facemash Creator Survives Ad Board, Harvard Crimson (Nov. 19, 2003), 23 https://www.thecrimson.com/article/2003/11/19/facemash-creator-survives-ad-board-the/; Bari Schwartz, Hot or Not? Website Briefly Judges Looks, Harvard Crimson (Nov. 4, 2003), 24 https://www.thecrimson.com/article/2003/11/4/hot-or-not-website-briefly-judges/; Sam Brodsky, Everything to Know About Facemash, the Site Zuckerberg Created in College to Rank 'Hot' 25 Women, Metro (Apr. 12, 2018), https://www.metro.us/everything-to-know-about-facemash-the- site-zuckerberg-created-in-college-to-rank-hot-women/; Noam Cohen (@noamcohen), Twitter 26 (Mar. 20, 2018, 3:27 PM). 197 The Social Network (Columbia Pictures 2010). 27 198 Richard MacManus, Facebook Mobile Usage Set to Explode, Read Write Web (Oct. 27 2011), 28 https://web.archive.org/web/20120520003847/http://www.readwriteweb.com/archives/facebook_ PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 52 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 59 of 278 1 Inc. had turned its initial losses into immense profitability, bringing in annual revenues of $3.7 2 billion and working with an operating income of $1.7 billion.199 3 188. But Facebook knew its future success was not guaranteed. On February 1, 2012, 4 Facebook, Inc. filed with the SEC for an initial public offering. The company's filing noted that its 5 historic performance might not continue indefinitely: "A number of other social networking 6 companies that achieved early popularity have since seen their active user bases or levels of 7 engagement decline, in some cases precipitously. There is no guarantee that we will not experience 8 a similar erosion of our active user base or engagement levels. A decrease in user retention, growth, 9 or engagement could render Facebook less attractive to developers and advertisers, which may have 10 a material and adverse impact on our revenue, business, financial condition, and results of 11 operations."200 12 189. Facebook, Inc. also disclosed that the proliferation of smartphones could materially 13 affect its ongoing prospects. "[O]ur users could decide to increasingly access our products primarily 14 through mobile devices. We do not currently directly generate any meaningful revenue from the 15 use of Facebook mobile products, and our ability to do so successfully is unproven. Accordingly, 16 if users continue to increasingly access Facebook mobile products as a substitute for access through 17 personal computers, and if we are unable to successfully implement monetization strategies for our 18 mobile users, our revenue and financial results may be negatively affected." 19 190. Facebook actively pursued changes to its product, including adding design features 20 offered to the public. As a result of these actions, Facebook achieved its goal. As of October 2021, 21 Facebook had ballooned to roughly 2.91 billion monthly active users, thus reaching 59% of the 22 23 24 25 mobile_usage_set_to_explode.php; Athima Chansanchai, One Third of Year's Digital Photos Are on Facebook, NBC News (Sept. 20, 2011), https://www.nbcnews.com/news/world/one-third- 26 years-digital-photos-are-facebook-flna120576. 199 Erick Schonfeld, Facebook's Profits: $1 Billion, On #3.7 Billion in Revenues, TechCrunch 27 (Feb. 1, 2012), https://techcrunch.com/2012/02/01/facebook-1-billion-profit/. 200 28 Facebook, Inc., Registration Statement (Form S-1) (Feb. 1, 2012) at 11, https://www.sec.gov/Archives/edgar/data/1326801/000119312512034517/d287954ds1.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 53 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 60 of 278 1 world's social networking population, the only social media product to reach over half of all social 2 media users. At least 6% of these users are children in the U.S. between the ages of 9 and 11. 201 3 191. Since its inception, Facebook has implemented several changes, developments, and 4 designs to its product to prolong user engagement and impose alterations to the user experience. As 5 discussed further below, several changes, developments, and designs render the product defective 6 and harmful. 7 b. Modifications of Facebook's product features over time. 8 192. When Meta launched thefacebook.com on February 4, 2004, only Harvard students 9 could create accounts using their university-issued email addresses. In March 2004, students at 10 Stanford, Columbia, and Yale were permitted to join, and eventually, any student with a college- 11 or university-issued email address could join Facebook. 202 In 2005, Facebook was opened to high 12 school students, provided they were invited by someone who was already using the site. 203 By 13 September 2006, Facebook was opened to all users.204 At the time, Meta claimed that it was open 14 only to persons aged 13 and older with a valid email address. 205 However, Meta did not require 15 verification of a user's age or identity and did not verify users' email addresses. As a result, 16 underage users could easily register an account with and access Facebook. 17 193. At first, Facebook was a collection of personal profiles and single photos. It was 18 described by the New York Times as "a fancy electronic version of the whiteboard that students 19 20 201 21 Katherine Schaeffer, 7 facts about Americans and Instagram, Pew Research Center (Oct. 7, 2021), https://www.pewresearch.org/fact-tank/2021/10/07/7-facts-about-americans-and- 22 instagram/. 202 Saul Hansell, Site Previously for Students Will Be Opened to Others, N.Y. Times (Sept. 12, 23 2006), https://www.nytimes.com/2006/09/12/technology/site-previously-for-students-will-be- opened-to-others.html. 24 203 Ellen Rosen, THE INTERNET; Facebook.com Goes to High School, N.Y. Times (Oct. 16, 2005), https://www.nytimes.com/2005/10/16/nyregion/the-internet-facebookcom-goes-to-high- 25 school.html. 204 Saul Hansell, Site Previously for Students Will Be Opened to Others, N.Y. Times (Sept. 12, 26 2006), https://www.nytimes.com/2006/09/12/technology/site-previously-for-students-will-be- opened-to-others.html. 27 205 Saul Hansell, Site Previously for Students Will Be Opened to Others, N.Y. Times (Sept. 12, 28 2006), https://www.nytimes.com/2006/09/12/technology/site-previously-for-students-will-be- opened-to-others.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 54 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 61 of 278 1 often mount on their doors to leave and receive messages." 206 Users could post a single profile 2 picture, add personal details such as gender, birthdate, phone number, and interests, or connect with 3 other users by "friending" them, either by searching for them or inviting them by email. Users could 4 also display their relationship statuses, or, alternatively what they were "[l]ooking for" (e.g., 5 friendship, dating, a relationship, "random play," or "whatever I can get") and "[i]nterested in" 6 (e.g., women, men). In September 2004, however, Meta introduced "the Wall," which allowed 7 users to interact with "friends" by posting on each other's profiles. This product feature kept users 8 returning to Facebook to monitor Wall activity. 9 194. In 2005, Facebook began allowing users to upload an unlimited amount of photos, 10 making it the first photo hosting website to do so. 207 11 195. In 2006, Meta introduced the Newsfeed to Facebook. 208 While previously "[e]very 12 browsing session was like a click-powered treasure hunt," 209 the Newsfeed provided a centralized 13 home page where users could view their friends' activity, including any changes to their profiles or 14 activity on the app, such as, for example, uploading new pictures, or a change in relationship 15 status.210 It was the first "social feed" of its kind, and increased time spent on the product. 211 Users 16 immediately decried this feature as an invasion of privacy. 212 Mark Zuckerberg rationalized the 17 feature by saying "we agree, stalking isn't cool; but being able to know what's going on in your 18 206 Ellen Rosen, THE INTERNET; Facebook.com Goes to High School, N.Y. Times (Oct. 16, 19 2005), https://www.nytimes.com/2005/10/16/nyregion/the-internet-facebookcom-goes-to-high- school.html. 20 207 Craig Kanalley, A History of Facebook Photos (Infographic), The Huffington Post (Aug. 2, 2011), https://www.huffpost.com/entry/facebook-photos-infographic_n_916225. 21 208 Think Marketing, This Is How Facebook Has Changed Over the Past 14 Years (February 6, 22 2018), https://thinkmarketingmagazine.com/facebook-celebrates-14-years-of-milestones-a- timeline/. 209 23 Jillian D'Onfro, Facebook's News Feed is 10 years old. This is how the site has changed, World Economic Forum (Sept. 9, 2016), https://www.weforum.org/agenda/2016/09/facebooks- 24 news-feed-is-10-years-old-this-is-how-the-site-has-changed. 210 Jillian D'Onfro, Facebook's News Feed is 10 years old. This is how the site has changed, 25 World Economic Forum (Sept. 9, 2016), https://www.weforum.org/agenda/2016/09/facebooks- news-feed-is-10-years-old-this-is-how-the-site-has-changed. 26 211 Jillian D'Onfro, Facebook's News Feed is 10 years old. This is how the site has changed, World Economic Forum (Sept. 9, 2016), https://www.weforum.org/agenda/2016/09/facebooks- 27 news-feed-is-10-years-old-this-is-how-the-site-has-changed. 212 28 Moneywatch, Facebook Under Fire for New Feature, CBS News (Sept. 7, 2006), https://www.cbsnews.com/news/facebook-under-fire-for-new-feature/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 55 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 62 of 278 1 friends' lives is."213 The Newsfeed algorithm was originally designed to maximize a user's time 2 spent in one session. However, Meta later changed the code to maximize as many use sessions as 3 possible. The frequency of sessions is a strong indicator of problematic use, a point internal 4 Facebook researchers have made when suggesting that Facebook should "help people consolidate 5 their use of Facebook into fewer sessions." 214 Despite this knowledge, Meta continued to focus on 6 maximizing sessions, including for teens, 215 even prioritizing the metric over "integrity" 7 improvements to its products.216 8 196. In May 2007, Meta launched a video service on Facebook, which allowed it to 9 compete with YouTube and the then-popular Myspace. 217 Users could upload videos, or record 10 them from a webcam. 11 197. In April 2008, Meta launched Facebook Chat, which later became Facebook 12 Messenger, allowing users to have private conversations with each other. 218 Facebook Chat 13 appeared as a permanent bar across the bottom of users' screens; it also provided users the ability 14 to see which friends were "online" and presumably available to chat. Facebook Chat allowed users 15 to immerse themselves even deeper into Meta's product; one commentator noted that "[b]y making 16 Facebook more real time/presence oriented, Facebook session length should go up a lot." 219 17 198. In May 2008, Meta added a "People You May Know" feature to the product, touting 18 it as a way to "connect [users] to more of your friends" on Facebook. 220 Facebook's algorithms 19 213 Gillian D'Onfro, Facebook's News Feed is 10 years old. This is how the site has changed, 20 World Economic Forum (Sept. 9, 2016), https://www.weforum.org/agenda/2016/09/facebooks- news-feed-is-10-years-old-this-is-how-the-site-has-changed. 21 214 Haugen_00010114 at Haugen_00010121. 215 See, e.g., META3047MDL-003-00161881 at META3047MDL-003-00161915 (highlighting 22 moderate decline in sessions among teen Instagram users in the United States) 216 23 See META3047MDL-003-00170806 at META3047MDL-003-00170822 (Instagram sessions "cannot decrease"). 217 24 Pete Cashmore, Facebook Video Launches: YouTube Beware!, Mashable (May 24, 2007), https://mashable.com/archive/facebook-video-launches. 25 218 Dan Farber, Facebook Chat begins to roll out, CNET (April 6, 2008), https://www.cnet.com/culture/facebook-chat-begins-to-roll-out/. 26 219 Dan Farber, Facebook Chat begins to roll out, CNET (April 6, 2008), https://www.cnet.com/culture/facebook-chat-begins-to-roll-out/. 27 220 Kashmir Hill, 'People You May Know:' A Controversial Facebook Feature's 10-Year History, 28 Gizmodo (Aug. 8, 2018), https://gizmodo.com/people-you-may-know-a-controversial-facebook- features-1827981959. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 56 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 63 of 278 1 utilize the vast amount of data it collects from its users to suggest users for "friending" to each 2 other.221 It utilizes information such as a user's friends list, their friends' friends list, education 3 information, and work information, along with other data, to make these suggestions. 222 Some users 4 dislike the feature, complaining that it constantly shows them people they do not want to friend, or 5 even suggests people in sexually explicit poses,223 but Facebook does not provide the option to 6 disable this feature. 7 199. In February 2009, Meta launched the "Like" button on Facebook. 224 The button 8 allowed users to quickly react to content, as opposed to typing out a comment. Facebook's 9 algorithm counts and displays likes to other users. The measure also served as a social measuring 10 stick, by which users could gauge the success of their posts, photographs, and videos. Soon after, 11 Meta expanded the "Like" feature to comments as well. Users could also use the "Like" button to 12 follow public figures, such as brands or publishers. When a user liked a brand, for example, Meta 13 would use that information to show ads for that brand to the user's friends on Facebook. 225 In April 14 2010, Meta launched "social plug-ins" that would allow people to "Like" things on the Internet 15 outside of Facebook. Meta used the button to track Facebook users' engagement across the Internet, 16 leveraging the data it gathered to target ads and fuel the Newsfeed algorithm. 226 The button also 17 221 Kashmir Hill, 'People You May Know:' A Controversial Facebook Feature's 10-Year History, 18 Gizmodo (Aug. 8, 2018), https://gizmodo.com/people-you-may-know-a-controversial-facebook- features-1827981959. 19 222 Kashmir Hill, 'People You May Know:' A Controversial Facebook Feature's 10-Year History, 20 Gizmodo (Aug. 8, 2018), https://gizmodo.com/people-you-may-know-a-controversial-facebook- features-1827981959. 223 21 Kashmir Hill, 'People You May Know:' A Controversial Facebook Feature's 10-Year History, Gizmodo (Aug. 8, 2018), https://gizmodo.com/people-you-may-know-a-controversial-facebook- 22 features-1827981959. 224 Will Oremus, How Facebook Designed the Like Button—and made social media into a 23 Popularity Contest, Fast Company (Nov. 15, 2022), https://www.fastcompany.com/90780140/the-inside-story-of-how-facebook-designed-the-like- 24 button-and-made-social-media-into-a-popularity-contest. 225 Will Oremus, How Facebook Designed the Like Button—and made social media into a 25 Popularity Contest, Fast Company (Nov. 15, 2022), https://www.fastcompany.com/90780140/the-inside-story-of-how-facebook-designed-the-like- 26 button-and-made-social-media-into-a-popularity-contest. 226 27 Will Oremus, How Facebook Designed the Like Button—and made social media into a Popularity Contest, Fast Company (Nov. 15, 2022), 28 https://www.fastcompany.com/90780140/the-inside-story-of-how-facebook-designed-the-like- button-and-made-social-media-into-a-popularity-contest. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 57 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 64 of 278 1 shaped users' own behavior, as they were conditioned to act and interact in whatever ways would 2 generate the like rewards, or risk having their content hidden from their friends' newsfeeds. 227 3 200. 2009 also marked the change from chronological to algorithmic ordering for the 4 news feed, with Meta now dictating which posts users would see by highlighting "Top Stories" in 5 each user's News Feed.228 6 201. In December 2010, Meta began using facial recognition to identify people in users' 7 Facebook photos and suggest that users tag them.229 Rather than letting users opt-in to the feature, 8 Meta automatically enabled it for all users. 230 9 202. Meta also debuted infinite scrolling in 2010, initially for photos specifically, but 10 later for its core News Feed, ensuring that users would never reach the bottom of a page and would, 11 instead, keep scrolling without end or limits, leading to excessive and compulsive product use. 231 12 203. In August 2012, Meta introduced the Facebook Messenger app, a feature that 13 allowed users to see when their friends were last active on the product. 232 14 15 16 17 18 227 Will Oremus, How Facebook Designed the Like Button—and made social media into a 19 Popularity Contest, Fast Company (Nov. 15, 2022), https://www.fastcompany.com/90780140/the-inside-story-of-how-facebook-designed-the-like- 20 button-and-made-social-media-into-a-popularity-contest. 228 21 Alex Hern, Facebook to Restore Chronological Feed of Posts from Friends, The Guardian (July 21, 2022), https://www.theguardian.com/technology/2022/jul/21/facebook-to-restore- 22 chronological-feed-of-posts-from-friends. 229 Ben Parr, Facebook Brings Facial Recognition to Photo Tagging, Mashable (Dec. 16, 2010), 23 http://www.cnn.com/2010/TECH/social.media/12/16/facebook.facial.recognition.mashable/index. html. 24 230 Charles Arthur, Facebook In New Privacy Row Over Facial Recognition Feature, The Guardian (June 8, 2011), https://www.theguardian.com/technology/2011/jun/08/facebook- 25 privacy-facial-recognition?INTCMP=SRCH. 231 Bob Leggit, How the Internet Destroyed Your Attention Span, Popzazzle (Apr. 30, 2021), 26 https://popzazzle.blogspot.com/2021/04/how-the-internet-destroyed-your-attention-span.html. 232 27 Billy Gallagher, Facebook Brings Notifications, Album-Specific Uploads to Standalone Camera App, Tech Crunch (Aug. 28, 2012), https://techcrunch.com/2012/08/28/facebook-brings- 28 notifications-album-specific-uploads-to-standalone-camera-app/?icid=tc_dan- schawbel_art&blogger=dan-schawbel#. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 58 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 65 of 278 1 204. In August 2015, Meta launched Facebook Live, which allowed users to live-stream 2 videos.233 It immediately prompted more engagement with the platform and furthered Meta's goal 3 of keeping users coming back, both to create the videos and to interact with them. 234 4 205. In February 2016, Meta expanded Facebook's "Like" feature for posts, adding 5 "Reactions" such as "like," "love," "haha," "wow," "sad," and "angry." 235 The following year, 6 reactions were extended to comments.236 In a manner similar to likes, these reactions further 7 manipulated adolescents' behavior, thus impacting their mental health and well-being and causing 8 damage and harm to certain Plaintiffs herein. 9 206. In March 2017, following the launch of a similar product on Instagram, Meta 10 introduced Facebook Stories, with the hope of competing with the success of Snapchat among 11 young people.237 With Stories, users could post short, ephemeral videos that appeared for 24-hours 12 at the top of friends' News Feeds. 238 Stories is designed to keep users coming back to the platform 13 at least daily, feeding performance metrics that are crucial to Meta's bottom line, or otherwise risk 14 missing out. 15 207. Later that year, in December 2017, Meta rolled out Facebook Kids, a messaging app 16 designed for kid's ages 6 to 12,239 for the purpose of getting younger users on its product sooner. 17 The app does not require a Facebook account, since COPPA prohibits users under the age of 13 18 19 233 Joe Lazauskus, The Untold Story of Facebook Live, Fast Company (Sept. 29, 2016), https://www.fastcompany.com/3064182/the-untold-story-of-facebook-live. 20 234 Joe Lazauskus, The Untold Story of Facebook Live, Fast Company (Sept. 29, 2016), https://www.fastcompany.com/3064182/the-untold-story-of-facebook-live. 21 235 Casey Newton, Facebook Rolls Out Expanded Like Button Reactions Around the World, The 22 Verge (Feb. 24, 2016), https://www.theverge.com/2016/2/24/11094374/facebook-reactions-like- button. 236 23 Natt Garun, Facebook Reactions Have Now Infiltrated Comments, The Verge (May 3, 2017), https://www.theverge.com/2017/5/3/15536812/facebook-reactions-now-available-comments. 24 237 Casey Newton, Facebook Launches Stories to Complete its All-out Assault on Snapchat, The Verge (Mar. 28, 2017), https://www.theverge.com/2017/3/28/15081398/facebook-stories- 25 snapchat-camera-direct. 238 Casey Newton, Facebook Launches Stories to Complete its All-out Assault on Snapchat, The 26 Verge (Mar. 28, 2017), https://www.theverge.com/2017/3/28/15081398/facebook-stories- snapchat-camera-direct. 27 239 Nick Statt, Facebook Launches a Version of Messenger for Young Children, The Verge (Dec. 28 4, 2017), https://www.theverge.com/2017/12/4/16725494/facebook-messenger-kids-app-launch- ios-iphone-preview. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 59 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 66 of 278 1 from signing up for such accounts, and instead allows children to create accounts that are managed 2 through parents' Facebook accounts. 240 Meta touted it as a way to "give[] parents more control." 241 3 The app, however, still collects an extraordinary amount of data about its child users, including the 4 content of their messages, any photos they send, and what features they use on the app. 242 Currently, 5 there are no other official Facebook products marketed publicly by Meta as intended for children 6 under 13 (despite the proliferation of such users on Instagram and Facebook). However, as of April 7 2021 Meta was actively seeking to develop ways for children as young as 6 to use the product. 243 8 208. In August 2020, Meta introduced "Reels" on Instagram. 244 Reels are short videos 9 posted by other Instagram users, presented in an algorithmically generated feed, and in a full-screen 10 format popularized by TikTok. Meta subsequently introduced Reels to Facebook in 2021. 245 As 11 explained more fully below, Meta committed to making videos more and more a part of their 12 platforms to attract and keep younger users in the face of competition from TikTok. 13 c. Facebook's acquisition and control of Instagram. 14 209. On or around April 6, 2012, Zuckerberg called Kevin Systrom, one of the co- 15 founders of Instagram, offering to purchase his company. 246 16 17 18 240 Nick Statt, Facebook Launches a Version of Messenger for Young Children, The Verge (Dec. 19 4, 2017), https://www.theverge.com/2017/12/4/16725494/facebook-messenger-kids-app-launch- ios-iphone-preview. 20 241 Loren Chang, Introducing Messenger Kids, a New App for Families to Connect, Meta (Dec. 4, 2017), https://about.fb.com/news/2017/12/introducing-messenger-kids-a-new-app-for-families-to- 21 connect/. 242 22 Nitasha Tiku, Facebook for 6-Year-Olds? Welcome to Messenger Kids, Wired (Dec. 5, 2017), https://www.wired.com/story/facebook-for-6-year-olds-welcome-to-messenger-kids/. 243 23 Ezra Kaplan and Jo Ling Kent, Documents reveal Facebook targeted children as young as 6 for consumer base, NBC News (Oct. 29, 2021), https://www.nbcnews.com/tech/social- 24 media/facebook-documents-reveal-company-targeted-children-young-6- rcna4021?cid=sm_npd_nn_tw_ma. 25 244 Instagram, Introducing Instagram Reels (Aug. 5, 2020), https://about.fb.com/news/2020/08/introducing-instagram-reels/. 26 245 Facebook, Launching Reels on Facebook in the US (Sept. 29, 2021), https://about.fb.com/news/2021/09/launching-reels-on-facebook-us/. 27 246 Nicholas Carlson, Here's The Chart That Scared Zuckerberg Into Spending $1 Billion On 28 Instagram, Insider (Apr. 14, 2012), https://www.businessinsider.com/heres-the-chart-that-scared- zuckerberg-into-spending-1-billion-on-instagram-2012-4. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 60 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 67 of 278 1 210. Instagram launched as a mobile-only app that allowed users to create, filter, and 2 share photos. On the first day of its release in October 2010, it gained a staggering 25,000 users. 247 3 By April 2012, Instagram had approximately 27 million users. When Instagram released an 4 Android version of its app—right around the time of Zuckerberg's call—it was downloaded more 5 than a million times in less than a day.248 Instagram's popularity is so widespread and image-based, 6 a new term has grown up around it for the perfect image or place: "Instagrammable." 249 7 211. On April 9, 2012, just days after Zuckerberg's overture to Systrom, Facebook, Inc. 8 purchased Instagram, Inc. for $1 billion in cash and stock. This purchase price was double the 9 valuation of Instagram implied by a round of funding the company closed days earlier. 250 10 212. Facebook, Inc. held its initial public offering less than two months after acquiring 11 Instagram, Inc.251 12 213. Zuckerberg's willingness to pay a premium for Instagram was driven by his instinct 13 that Instagram would be vital to reaching a younger, smartphone-oriented audience—and thus 14 critical to his company's going-forward success. 15 214. This was prescient. Instagram's revenue grew exponentially from 2015 to 2022. 252 16 A study conducted in the second quarter of 2018 showed that, over the prior year, advertisers' 17 spending on Instagram grew by 177%—more than four times the growth of ad spending on 18 19 247 Dan Blystone, Instagram: What It Is, Its History, and How the Popular App Works, Investopedia (Oct. 22, 2022), https://www.investopedia.com/articles/investing/102615/story- 20 instagram-rise-1-photo0sharing-app.asp#:~:text=History%20of%20Instagram. 248 Kim-Mai Cutler, From 0 to $1 billion in two years: Instagram's rose-tinted ride to glory 21 TechCrunch (Apr. 9, 2012), https://techcrunch.com/2012/04/09/instagram-story-facebook- acquisition/. 22 249 Sarah Frier, No Filter 81 (2020). 23 250 Alexia Tsotsis, Right Before Acquisition, Instagram Closed $50M At A $500M Valuation From Sequoia, Thrive, Greylock And Benchmark, TechCrunch (Apr. 9, 2012), 24 https://techcrunch.com/2012/04/09/right-before-acquisition-instagram-closed-50m-at-a-500m- valuation-from-sequoia-thrive-greylock-and-benchmark/. 25 251 Evelyn Rusli & Peter Eavis, Facebook Raises $16Billion in I.P.O., N.Y. Times (May 17, 2012), https://archive.nytimes.com/dealbook.nytimes.com/2012/05/17/facebook-raises-16-billion- 26 in-i-p-o/. 252 27 See Josh Constine, Instagram Hits 1 Billion Monthly Users, Up From 800M in September, TechCrunch (June 20, 2018), https://techcrunch.com/2018/06/20/instagram-1-billion-users/ 28 (showing meteoric rise in monthly active users over period and reporting year-over-year revenue increase of 70% from 2017-2018). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 61 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 68 of 278 1 Facebook.253 Likewise, visits to Instagram rose by 236%, nearly thirty times the growth in site visits 2 experienced by Facebook during the same period.254 By 2021, Instagram accounted for over half 3 of Meta's $50.3 billion in net advertising revenues. 255 4 215. Meta has claimed credit for Instagram's success since its acquisition. Zuckerberg 5 told market analysts that Instagram "wouldn't be what it is without everything that we put into it, 6 whether that's the infrastructure or our advertising model."256 7 216. Instagram has become the most popular photo-sharing social media product among 8 teenagers and young adults in the United States. 62% of American teens use Instagram, with 10% 9 of users reporting that they use it "almost constantly."257 Instagram's young user base has become 10 even more important to Meta as the number of teens using Facebook has decreased over time. 258 11 217. Facebook's and Instagram's success, and the riches they have generated for Meta, 12 have come at an unconscionable cost in human suffering. In September 2021, The Wall Street 13 Journal began publishing internal documents leaked by former Facebook product manager Frances 14 Haugen.259 15 218. The documents are disturbing. They reveal that, according to Meta's researchers, 16 13.5% of U.K. girls reported more frequent suicidal thoughts and 17% of teen girls reported 17 253 Merkle, Digital Marketing Report 3 (Q2 2018), https://www.merkleinc.com/thought- 18 leadership/digital-marketing-report/digital-marketing-report-q2-2018. 254 Merkle, Digital Marketing Report 19 (Q2 2018), https://www.merkleinc.com/thought- 19 leadership/digital-marketing-report/digital-marketing-report-q2-2018. 255 Sara Lebow, For the First Time, Instagram Contributes Over Half of Facebook's US Ad 20 Revenues, eMarketer (Nov. 2, 2021), https://www.emarketer.com/content/instagram-contributes- over-half-of-facebook-us-ad-revenues. 21 256 Salvador Rodriguez, Mark Zuckerberg Is Adamant that Instagram Should Not Be Broken Off 22 from Facebook, CNBC (Oct. 20, 2019), https://www.cnbc.com/2019/10/30/mark-zuckerberg-is- adamant-that-instagram-should-remain-with-facebook.html. 23 257 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Research. Ctr. (Aug. 10, 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 24 2022/; see also Piper Sandler, Taking Stock With Teens 19 (Fall 2021), https://piper2.bluematrix.com/docs/pdf/3bad99c6-e44a-4424-8fb1- 25 0e3adfcbd1d4.pdf?utm_source=newsletter&utm_medium=email&utm_campaign=newsletter_axi osam&stream=top (eighty-one percent of teens use Instagram at least once a month). 26 258 Sheera Frenkel et al., Instagram Struggles with Fears of Losing Its 'Pipeline': Young Users, N.Y. Times (Oct. 26, 2021), available at 27 https://www.nytimes.com/2021/10/16/technology/instagram-teens.html. 259 28 The collection of Wall Street Journal articles are available online via the following link: https://www.wsj.com/articles/the-facebook-files-11631713039?mod=bigtop-breadcrumb. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 62 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 69 of 278 1 worsening eating disorders after starting to use Instagram. 260 Over 40% of Instagram users who 2 reported feeling "unattractive" said that feeling began while using Instagram, 261 and 32% of teen 3 girls who already felt bad about their bodies felt even worse because of the app. 262 4 219. Internal Meta presentations from 2019 and 2020 were unsparing in their conclusions 5 about the harms caused by Instagram: "We make body image issues worse for one in three teen 6 girls." "Mental health outcomes related to this can be severe." "Aspects of Instagram exacerbate 7 each other to create a perfect storm."263 8 220. Haugen's revelations made clear to the public what Meta has long known: in an 9 effort to addicts kids and promote usage, Meta's products exploit the neurobiology of developing 10 brains, and all the insecurities, status anxieties, and beauty comparisons that come along with it. In 11 a bid for higher profits, Meta ignored the harms resulting from its overuse-oriented business model, 12 which are widespread, serious, long-term, and in tragic instances fatal. 13 d. Modifications of Instagram's product features over time. 14 221. In its earliest form, Instagram was a photo-sharing app. Users could post still 15 images—enhanced by the product's suite of built-in photo filters—"follow" other users, and "Like" 16 or comment on posts by other users, all in a centralized chronological feed. Instagram also allowed 17 18 260 Morgan Keith, Facebook's Internal Research Found its Instagram Platform Contributes to 19 Eating Disorders and Suicidal Thoughts in Teenage Girls, Whistleblower Says, Insider (Oct. 3, 2021), https://www.businessinsider.com/facebook-knows-data-instagram-eating-disorders- 20 suicidal-thoughts-whistleblower-2021-10. 261 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for 21 Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- 22 documents-show-11631620739; Facebook Staff, Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the U.S. 9 (Mar. 26, 2020), 23 https://s.wsj.net/public/resources/documents/teen-girls-body-image-and-social-comparison-on- instagram.pdf. 24 262 Billy Perrigo, Instagram Makes Teen Girls Hate Themselves. Is That a Bug or a Feature?, Time (Sept. 16, 2021), https://time.com/6098771/instagram-body-image-teen-girls/. 25 263 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for 26 Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- 27 documents-show-11631620739; Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the U.S., Wall St. J. (Sept. 29, 2021), 28 https://s.wsj.net/public/resources/documents/teen-girls-body-image-and-social-comparison-on- instagram.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 63 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 70 of 278 1 users to see their friends' activity—such as liking or commenting on a post, or following other 2 accounts—on the app, through its "Following" tab. 3 222. In January 2011, Instagram added hashtags, which allowed users to group together 4 posts about particular topics.264 5 223. Since acquiring Instagram, Meta has introduced to the product a host of additional 6 features to drive pre-teen and teenage engagement and, in doing so, increase advertising revenues. 7 224. In June 2013, in addition to the still, filtered images for which the product was 8 known, Instagram began to support videos of up to 15 seconds. 265 This feature also included 13 9 new specially-created filters that could be applied to the videos. At the time, this feature satisfied 10 what some characterized as the "years-long search for an 'Instagram for video,'" 266 and allowed 11 Instagram to compete with a popular video-sharing product at the time, Vine. It also allowed users 12 posting videos to select their "favorite" scene from the video to be displayed as the cover image on 13 video posts. According to Systrom, this ensured that user's videos were "beautiful even when 14 they're not playing."267 Despite this, Instagram remained largely a photo-sharing app. 15 225. In December 2013, Meta added Instagram Direct, a feature that allows users to send 16 messages or posts to specific people directly from the app. 268 This function allowed Instagram to 17 compete against messaging products like Snapchat that were gaining popularity among teens and 18 pre-teens. 19 20 21 22 264 Diving Deep into the Science of the Instagram Algorithm, Signalytics, 23 https://about.fb.com/news/2020/08/introducing-instagram-reels/. 265 Colleen Taylor, Instagram Launches 15-Second Video Sharing Feature, With 13 Filters And 24 Editing, Tech Crunch (June 20, 2013), https://techcrunch.com/2013/06/20/facebook-instagram- video/. 25 266 Colleen Taylor, Instagram Launches 15-Second Video Sharing Feature, With 13 Filters And Editing, Tech Crunch (June 20, 2013), https://techcrunch.com/2013/06/20/facebook-instagram- 26 video/. 267 Kevin Systrom, Introducing Video on Instagram, Instagram (June 20, 2013), 27 https://about.instagram.com/blog/announcements/introducing-video-on-instagram. 268 28 Jordan Crook, Instagram Introduces Instagram Direct, Tech Crunch (Dec. 12, 2013), https://techcrunch.com/2013/12/12/instagram-messaging/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 64 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 71 of 278 1 226. In June 2015, Meta opened Instagram to all advertisers, weaving advertisements into 2 users' Feeds.269 3 227. In March 2016, Meta did away with Instagram's chronological feed and 4 implemented engagement-based ranking algorithms.270 Now, upon opening the app, users would 5 no longer see posts from people they followed in the order they were posted; instead, they would 6 encounter an algorithmic feed, similar to the one employed on Meta's other product, Facebook. At 7 the time, Meta said that the new algorithmic feed would rank the order of posts in users' feeds 8 based on the "likelihood that [they would] be interested in the content, [their] relationship with the 9 person posting and the timeliness of the post. . . ." 271 10 228. In February 2016, with the popularity of video content rising on Instagram, Meta 11 added view counts to videos, allowing users to see how many times users had viewed their posts. 272 12 Later that year, in December 2016, Instagram added the ability to "Like" comments to posts 13 (symbolized by a heart emoji).273 Both features became a source of additional motivation by users 14 to seek social acceptance and validation. 15 229. In August 2016, Meta introduced Instagram Stories, 274 another feature intended to 16 compete against Snapchat for the youth market. Systrom has admitted that the feature was copied 17 18 19 269 Vindu Goel & Sydney Ember, Instagram to Open its Photo Feed to Ads, N.Y. Times (June 2, 20 2015), https://www.nytimes.com/2015/06/03/technology/instagram-to-announce-plans-to- expand-advertising.html. 21 270 Alex Heath, Instagram is about to go through its most radical change ever, Insider (Mar. 15, 22 2016), https://www.businessinsider.com/instagram-to-introduce-algorithmic-timeline-2016-3. 271 Alex Heath, Instagram is about to go through its most radical change ever, Insider (Mar. 15, 23 2016), https://www.businessinsider.com/instagram-to-introduce-algorithmic-timeline-2016-3. 272 Michael Zhang, Instagram is Adding View Counts to Your Videos, PetaPixel (Feb. 11, 2016), 24 https://petapixel.com/2016/02/11/instagram-adding-view-counts-videos/. 273 Hayley Tsukayama, Instagram will soon let you like comments – or even turn them off 25 completely, Wash. Post (Dec. 6, 2016), https://www.washingtonpost.com/news/the- switch/wp/2016/12/06/instagram-will-soon-let-you-like-comments-or-even-turn-them-off- 26 completely/. 274 27 Instagram, Introducing Instagram Stories (Aug. 2, 2016), https://about.instagram.com/blog/announcements/introducing-instagram- 28 stories#:~:text=Today%2C%20we're%20introducing%20Instagram,a%20slideshow%20format% 3A%20your%20story. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 65 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 72 of 278 1 from a Snapchat feature popular with children called "Snapchat Stories."275 Later that year, in 2 November 2016, Meta introduced another feature, Instagram Live, 276 designed to compete with 3 both Snapchat's ephemeral, disappearing posts, and the live-streamed videos of a then-popular 4 product called Periscope. Live permitted users to live stream video, which disappeared as soon as 5 the live stream stopped. 6 230. In December 2016, Meta introduced a product feature that allowed users to "save" 7 posts from other users.277 By tapping a bookmark icon underneath posts in their feeds, users could 8 save posts for later, in a private tab that was viewable only to the saving user. 9 231. In April 2017, Meta introduced another feature with appeal to children, an offline 10 mode that allows users to view posts and interact with Instagram even when they do not have access 11 to an Internet connection,278 for example when riding a bus to or from school. 12 232. In January 2018, Meta launched a feature allowing Instagram users to see when 13 others they had messaged with were active, or most recently active, on Instagram. This feature 14 exploits social reciprocity which, as explained above, results in more time spent using the product. 15 233. In June 2018, at the same time it announced that Instagram had grown to one billion 16 users, Meta introduced IGTV, both in the Instagram app and as a standalone product. 279 IGTV was 17 intended to rival YouTube. IGTV allowed users to upload videos up to one-hour long. 18 234. In September 2018, Systrom and Instagram co-founder Mike Krieger resigned from 19 Instagram, and Facebook named Adam Mosseri, a 10-year veteran of Facebook, as Instagram's 20 new CEO. 21 22 275 Rachel Kraus, Instagram Founder Admits He Blatantly Stole Stories from Snapchat, Mashable 23 (Mar. 11, 2019), https://mashable.com/article/kevin-systrom-instagram-stories-snapchat; Sheera Frenkel et al., Instagram Struggles With Fears of Losing Its 'Pipeline': Young Users, N.Y. Times 24 (Oct. 16, 2021), https://www.nytimes.com/2021/10/16/technology/instagram-teens.html. 276 Josh Constine, Instagram launches disappearing Live video and messages, Tech Crunch (Nov. 25 21, 2016), https://techcrunch.com/2016/11/21/instagram-live/. 277 Lisa Eadicicco, Instagram Just Added a New Feature that Lets You Save Other Users' Posts, 26 Time (Dec. 14, 2016), https://time.com/4602063/instagram-new-update-features-2016/. 278 Josh Constine, Instagram on Android gets offline mode, Tech Crunch (Apr. 18, 2017), 27 https://techcrunch.com/2017/04/18/instagram-offline/. 279 28 Kevin Systrom, Welcome to IGTV, our New Video App, Instagram (June 20, 2018), https://about.instagram.com/blog/announcements/welcome-to-igtv. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 66 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 73 of 278 1 235. Under Mosseri's leadership, Instagram aggressively focused on acquiring and 2 maximizing the engagement of young users. In 2018, Instagram allotted most of its global annual 3 marketing budget to targeting 13- to 15-year old children, a marketing demographic it calls "early 4 high school."280 According to Meta, these users represent the platform's "teen foothold" for its "US 5 pipeline."281 "Youth and Teens are critically important to Instagram. While Instagram has strong 6 market-fit with Teens, we know we need to constantly 're-win' this segment." 282 Meta has expressly 7 sought to maximize metrics like "teen time spent" on the Instagram product. 283 8 236. One way Meta sought to increase its teen metrics was through its launch of "Reels" 9 in August 2020, which mimicked the format of videos on TikTok. As noted, Reels mirrors TikTok 10 by algorithmically presenting short, "full-screen" videos posted by other Instagram users. Like 11 TikTok, Reels counts and displays the number of likes, follows, comments, shares, and views of a 12 video. The following year, Meta did away with IGTV, which had allowed longer videos to be 13 posted by users in a more traditional format. In late July 2022, Mosseri announced that "more and 14 more of Instagram is going to become video over time." 284 15 237. Instagram creates images and GIFs for users to incorporate into their videos and 16 picture postings. Instagram has also acquired publishing rights to thousands of hours of music and 17 video which it provides to its users to attach to the videos and pictures that they post on Instagram. 18 19 20 21 280 Sheera Frenkel et al., Instagram Struggles With Fears of Losing Its 'Pipeline': Young Users, 22 N.Y. Times (Oct. 16, 2021), https://www.nytimes.com/2021/10/16/technology/instagram- teens.html. 281 23 Sheera Frenkel et al., Instagram Struggles With Fears of Losing Its 'Pipeline': Young Users, N.Y. Times (Oct. 16, 2021), https://www.nytimes.com/2021/10/16/technology/instagram- 24 teens.html. 282 META3047MDL-003-00030070 at META3047MDL-003-00030071. 25 283 Sheera Frenkel et al., Instagram Struggles with Fears of Losing Its 'Pipeline': Young Users, N.Y. Times (Oct. 26, 2021), available 26 at https://www.nytimes.com/2021/10/16/technology/instagram-teens.html. 284 27 Marisa Dellatto, Instagram Exec Defends Shift to Video Despite Complaints from Creators like Kylie Jenner, Forbes (July 26, 2022), 28 https://www.forbes.com/sites/marisadellatto/2022/07/26/instagram-exec-defends-shift-to-video- despite-complaints-from-creators-like-kylie-jenner/?sh=4099badd5c6e. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 67 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 74 of 278 1 2. Meta intentionally encourages youth to use its products and then leverages that usage to increase revenue. 2 3 238. Facebook and Instagram owe their success to their defective design, including their 4 underlying computer code and algorithms, and to Meta's failure to warn Plaintiffs and Consortium 5 Plaintiffs that the products present serious safety risks. Meta's tortious conduct begins before a user 6 has viewed, let alone posted, a single scrap of content. 7 239. Meta describes the Instagram product as a "mobile-first experience." 285 Indeed, the 8 great majority of Instagram users in the U.S. access Instagram through a mobile application for 9 either the iOS or Android operating system. 10 240. In order to use the Facebook or Instagram app, one must first obtain it. On a mobile 11 device, this is accomplished by visiting a store from which the product can be downloaded—either 12 the Apple App Store (for iPhone users) or the Google Play Store (for Android users). Once installed 13 onto an individual's smartphone, they can open the app. They are then asked to create a new account 14 by entering an email address, adding a name, and creating a password and username. 15 241. A prospective Instagram or Facebook user is then invited to press a colorful button 16 that says "Sign up." In small print above this button, the user is informed: "By tapping Sign up, 17 you agree to our Terms, Data Policy and Cookies Policy." The text of those policies is not presented 18 on the sign-up page. While the words "Terms," "Data Policy," and "Cookies Policy" are slightly 19 bolded, the user is not informed that they can or should click on them, or otherwise told how they 20 can access the policies. 21 22 23 24 25 26 27 285 28 Yorgos Askalidis, Launching Instagram Messaging on Desktop, Instagram (Sept. 25, 2020), https://about.instagram.com/blog/engineering/launching-instagram-messaging-on-desktop. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 68 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 75 of 278 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 242. Meta's Data Policy (rebranded as a "Privacy Policy" in 2022), which applies to a 21 raft of Meta apps, including Facebook and Instagram,286 indicates Meta collects a breathtaking 22 amount of data from the users of its products, including: 23 "[c]ontent that you create, such as posts, comments or audio;" 24 "[c]ontent you provide through our camera feature or your camera roll 25 settings, or through our voice-enabled features;" 26 27 286 28 Meta, Privacy Policy, Meta (Jan. 1 2023), https://mbasic.facebook.com/privacy/policy/printable/#annotation-1. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 69 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 76 of 278 1 "[I]nformation you've shared with us through device settings, such as GPS 2 location, camera access, photos and related metadata;" 3 "[m]essages that you send and receive, including their content;" 4 "Metadata about content and messages;" 5 "[t]ypes of content that you view or interact with, and how you interact with 6 it;" 7 "[t]he time, frequency and duration of your activities on our products;" 8 "your contacts' information, such as their name and email address or phone 9 number, if you choose to upload or import it from a device, such as by 10 syncing an address book;" 11 information about "What you're doing on your device (such as whether our 12 app is in the foreground or if your mouse is moving);" 13 "device signals from different operating systems," including "things such as 14 nearby Bluetooth or Wi-Fi connections;" 15 "[i]nformation about the network that you connect your device to," which 16 includes "The name of your mobile operator or Internet service provider 17 (ISP), Language, Time zone, Mobile phone number, IP address, Connection 18 speed, Information about other devices that are nearby or on your network, 19 Wi-Fi hotspots you connect to using our products;" and 20 "information from . . . third parties, including . . . [m]arketing and advertising 21 vendors and data providers, who have the rights to provide us with your 22 information." 23 243. While the Data Policy indicates the scope of user information collected by Meta 24 through Facebook and Instagram, it is far less forthcoming about the purposes for which this data 25 is collected, and its consequences for younger users. 26 244. The Data Policy presents those goals as benign and even positive for its users—" to 27 provide a personalized experience to you" and to "make suggestions for you such as people you 28 may know, groups or events that you may be interested in or topics that you may want to follow." PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 70 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 77 of 278 1 245. The Data Policy does not inform users, and did not inform Plaintiffs, that the more 2 time individuals spend using Facebook and Instagram, the more ads Meta can deliver and the more 3 money it can make, or that the more time users spend on Facebook and Instagram, the more Meta 4 learns about them, and the more it can sell to advertisers the ability to micro-target highly 5 personalized ads.287 6 246. Meta monetizes its users and their data by selling ad placements to marketers. Meta 7 generated $69.7 billion from advertising in 2019, more than 98% of its total revenue for the year. 288 8 247. Given its business model, Meta has every incentive to—and knowingly does— 9 addict users to Facebook and Instagram. It accomplishes this through the algorithms that power its 10 apps, which are designed to induce compulsive and continuous scrolling for hours on end, operating 11 in conjunction with the other defective features described throughout this Complaint. 289 12 248. Meta's Data Policy contains no warnings whatsoever that use of its products at the 13 intensity and frequency targeted by Meta creates known risks of mental, emotional, and behavioral 14 problems for children, Instagram's key audience. 15 16 287 Nor does it inform users that Meta has allowed third-party apps to harvest from Facebook 17 "vast quantities of highly sensitive user and friends permissions." In re Facebook, Inc., No. 18- md-02843-VC, ECF No. 1104 at 9 (N.D. Cal. Feb. 9, 2023). This has included an app called 18 Sync.Me, which—according to Meta's internal investigative documents—"had access to many 'heavyweight' permissions," "including the user's entire newsfeed, friends' likes, friends' 19 statuses, and friends' hometowns." In re Facebook, Inc., No. 18-md-02843-VC, ECF No. 1104 at 9 (N.D. Cal. Feb. 9, 2023). It has included Microstrategy, Inc., which accessed data from "16 to 20 20 million" Facebook users, despite only being installed by 50,000 people. In re Facebook, Inc., No. 18-md-02843-VC, ECF No. 1104 at 9 (N.D. Cal. Feb. 9, 2023). And it has included one 21 Yahoo app that made "billions of requests" for Facebook user information, including "personal information about those users' friends, including the friends' education histories, work histories, 22 religions, politics, 'about me' sections, relationship details, and check-in posts." In re Facebook, Inc., No. 18-md-02843-VC, ECF No. 1104 at 9-10 (N.D. Cal. Feb. 9, 2023). 23 288 Rishi Iyengar, Here's How Big Facebook's Ad Business Really Is, CNN (July 1, 2020), https://www.cnn.com/2020/06/30/tech/facebook-ad-business-boycott. 24 289 See Christian Montag, et al., Addictive Features of Social Media/Messenger Platforms and 25 Freemium Games against the Background of Psychological and Economic Theories, 16 Int'l J. Env't Rsch. and Pub. Health 2612, 5 (July 2019), 26 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6679162/ ("One technique used to prolong usage time in this context is the endless scrolling/streaming feature."); see generally, Ludmila 27 Lupinacci, 'Absentmindedly scrolling through nothing': liveness and compulsory continuous connectedness in social media, 43 Media, Culture & Soc'y 273 (2021), 28 https://journals.sagepub.com/doi/epdf/10.1177/0163443720939454 (describing the ways that users use and experience social media apps). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 71 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 78 of 278 1 249. Instagram's collection and utilization of user data begins the instant a user presses 2 "Sign Up." At that point, Instagram prompts a new user to share a substantial amount of personal 3 data. First, Instagram asks the user to share their personal contacts, either by syncing contacts from 4 their phone and/or syncing their "Friends" from Facebook—"We'll use your contacts to help you 5 find your friends and help them find you." Next, Instagram asks the new user to upload a photo of 6 themselves. After that, Instagram asks the user to "Choose your interests" in order to "Get started 7 on Instagram with account recommendations tailored to you." And finally, Instagram invites the 8 new user to "Follow accounts to see their photos and videos in your feed," offering a variety of 9 recommendations. After sign-up is completed, Instagram prompts the new user to post either a 10 photo or a short video. 11 250. Meta's collection and utilization of user data continues unabated as a new user 12 begins to interact with its products. Meta's tracking of behavioral data—ranging from what the user 13 looks at, to how long they hover over certain images, to what advertisements they click on or 14 ignore—helps Meta build out a comprehensive and unique fingerprint of the user's identity. As the 15 user continues to use the product, Meta's algorithm works silently in the background to refine this 16 fingerprint, by continuously monitoring and measuring patterns in the user's behavior. Meta's 17 algorithm is sophisticated enough that it can leverage existing data to draw educated inferences 18 about even the user behavior it does not track firsthand. Meta's comprehensive data collection 19 allows it to target and influence its users in order to increase their "engagement" with its apps. 20 251. Meta's collection and analysis of user data allows it to assemble virtual dossiers on 21 its users, covering hundreds if not thousands of user-specific data segments. This, in turn, allows 22 advertisers to micro-target marketing and advertising dollars to very specific categories of users, 23 who can be segregated into pools or lists using Meta's data segments. Only a fraction of these data 24 segments come from content knowingly designated by users for publication or explicitly provided 25 by users in their account profiles. Many of these data segments are collected by Meta through covert 26 surveillance of each user's activity while using the product and when logged off the product, 27 including behavioral surveillance that users are unaware of, like navigation paths, watch time, and 28 hover time. Essentially, the larger Meta's user database grows, the more time the users spend on PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 72 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 79 of 278 1 the database, and the more detailed information that Meta can extract from its users, the more 2 money it makes. 3 252. Currently, advertisers can target Instagram and Facebook ads to young people based 4 on age, gender, and location.290 According to U.S.-based non-profit Fairplay, Meta did not actually 5 cease collecting data from teens for advertising in July 2021, as Meta has claimed. 291 6 253. Meta clearly understands the revenue and growth potential presented by its youngest 7 users, and it is desperate to retain them. Documents obtained by The New York Times indicate, that 8 since 2018, almost all of Instagram's $390 million global marketing budget has gone towards 9 showing ads to teenagers.292 10 254. Before the rise of Instagram, Facebook was the social media product by which Meta 11 targeted young users. Until recently, this targeting was devastatingly effective. In January 2014, 12 90% of U.S. teens used Facebook monthly; as late as January 2016, 68% did. 293 13 255. While the number of teen Facebook users has declined in recent years, Facebook 14 remains critical to Meta's strategy towards young users. Meta views Facebook as the nexus of teen 15 users' lives on social media, "where all social circles intersect," and as filling a similar role for such 16 users as the career-focused social media product LinkedIn fills for adults. 294 According to the 17 summary of a 2018 meeting, Meta's expressed goal was to have users "move through" Meta's 18 products "as they grow, i.e. Messenger Kids → Instagram → Facebook." 295 19 256. To create this cycle, and notwithstanding restrictions under COPPA, Meta embarked 20 on a "major investment in youth," researching and pursuing products targeted at kids as young as 21 22 290 Andrea Vittorio, Meta's Ad-Targeting to Teens Draws Advocacy Group Opposition, Bloomberg (Nov. 16, 2021), https://news.bloomberglaw.com/privacy-and-data-security/metas-ad- 23 targeting-to-teens-draws-advocacy-group-opposition. 291 24 Andrea Vittorio, Meta's Ad-Targeting to Teens Draws Advocacy Group Opposition, Bloomberg (Nov. 16, 2021), https://news.bloomberglaw.com/privacy-and-data-security/metas-ad- 25 targeting-to-teens-draws-advocacy-group-opposition. 292 Sheera Frenkel, et al, Instagram Struggles With Fears of Losing Its 'Pipeline': Young Users 26 N.Y. Times (Oct. 16, 2021), https://www.nytimes.com/2021/10/16/technology/instagram- teens.html. 27 293 META3047MDL-003-00171899 at META3047MDL-003-00171904. 294 META3047MDL-003-00171899 at META3047MDL-003-00171909. 28 295 META3047MDL-003-00003731 at META3047MDL-003-00003732. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 73 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 80 of 278 1 six.296 The centerpiece of these efforts is Messenger Kids ("MK"). 297 In 2019, Meta conducted at 2 least two research projects on growing MK. One study explored how to use "Playdates as a Growth 3 Lever for Messenger Kids."298 During this study, Meta sought to understand better how playdates 4 might be an area to increase usage among kids by interviewing parents of active users and the young 5 users themselves.299 Investigators suggested there was an opportunity to "brainstorm features 6 and/or prompts encouraging use of the app prior and after playdates to improve retention and active 7 threads."300 Later that year, they released a finding from a second investigation of parents and 8 children who used MK and those who did not. 301 To drive MK growth, the study recommended 9 "encourag[ing] more K2K [kid-to-kid] connections in MK" by "surfac[ing] and develop[ing] 10 additional in-app activities that involve others," while emphasizing to parents the "play-based 11 messaging" and the "play aspect of MK—camera filters, games, filters via notifs and QPs." 302 12 These are many of the same defective features found in Instagram. 13 257. Meta was also eager to market its products to tweens—users aged 10-12. Although 14 Meta employees publicly denied using children as "guinea pigs" to develop product features, 15 internally Meta was intensely interested in children's use of their apps. 303 It conducted research 16 projects with titles such as "Tweens JTBD Survey" 304 and "Exploring Tweens Social Media 17 Habits."305 In the latter study, Meta compared tween perceptions of their competitors' products to 18 understand "tween product needs,"306 noting that tweens can "connect and have fun using existing 19 296 Haugen_00017238 at Haugen_00017238. 20 297 Nick Stat, Facebook launches a version of Messenger for young children, The Verge (December 4, 2022) https://www.theverge.com/2017/12/4/16725494/facebook-messenger-kids- 21 app-launch-ios-iphone-preview. 298 22 Haugen_00023087 at Haugen_00023087. 299 Haugen_00023087 at Haugen_00023088, Haugen_00023097. 23 300 Haugen_00023087 at Haugen_00023090. 301 Haugen_00023066 at Haugen_00023066. 24 302 Haugen_00023066 at Haugen_00023085. 303 25 John Twomey, Molly Russell Inquest Latest: Teenager Viewed Suicide Videos of 'Most Distressing Nature', Express (Sept. 23, 2022), 26 https://www.express.co.uk/news/uk/1673461/Molly-Russell-inquest-latest-Teenager-suicide- videos-instagram. 27 304 "JTBD" appears to stand for "Jobs to Be Done." Haugen_00024450 at Haugen_00024454. 305 Haugen_00023849 at Haugen_00023850. 28 306 Haugen_00023849 at Haugen_00023888. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 74 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 81 of 278 1 apps, even though they're not made with a 10-to-12-year-old in mind." 307 Meta's takeaway was to 2 "use entertainment/interest as a starting point for engagement" and to "highlight fitting in." 308 3 258. In 2019, Meta conducted a series of interviews in Los Angeles and Denver with 4 tween friend groups, friend pairs, and individuals.309 Meta used this research to craft "product 5 recommendations" to appeal to tweens, suggesting features to help "decrease friction in the digital 6 interaction funnel."310 The recommendations included developing ways to "provide automatic 7 signals that indicate whether friends are available to interact," "[e]nable tweens to instrumentally 8 signal their availability," "[p]rovide light conversations starters that tweens can use to test the 9 reciprocity of an interaction (e.g., poking, waves)," and "build in a way that enables quick 10 communication across all messaging modalities."311 11 259. Meta's interest, efforts, and success in expanding the presence of its products in 12 children's lives is clear. Given the delicate, developing nature of the young brain and Meta's 13 creation of social media products designed to promote repetitive, compulsive use, it is not surprising 14 that American society is now grappling with the ramifications of Meta's growth-at-any-cost 15 approach. In a candid moment, a Software Engineer at Meta admitted, "It's not a secret that we've 16 often resorted to aggressive tactics in the name of growth, and we've been pretty unapologetic about 17 it."312 18 260. Meta has studied features and designs from its other products to make Instagram as 19 attractive and addictive as possible to young users. Meta's flagship product Facebook was the 20 original testing ground for many of Instagram's addicting and otherwise defective features, which 21 the two products share to this day. This feature overlap is no accident: it represents a conscious 22 strategy adopted by Meta to keep social media users hooked on its "family" of products for their 23 entire lives. 24 25 307 Haugen_00023849 at Haugen_00023886. 308 26 Haugen_00023849 at Haugen_00023888. 309 Haugen_00024450 at Haugen_00024450. 27 310 Haugen_00024450 at Haugen_00024466. 311 Haugen_00024450 at Haugen_00024466. 28 312 Haugen_00000934 at Haugen_00000934. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 75 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 82 of 278 1 261. From the beginning, both the Facebook and Instagram products have exploited 2 vulnerabilities in human psychology to addict users and maximize user time and engagement. 3 Facebook's first President, Sean Parker, summed up the devastating impact of this product design 4 in a 2017 interview: 5 God only knows what it's doing to our children's brains. . . . The thought process that went into building these applications, Facebook 6 being the first of them, . . . was all about: 'How do we consume as much of your time and conscious attention as possible?' . . . And that 7 means that we need to sort of give you a little dopamine hit every once in a while, because someone liked or commented on a photo or 8 a post . . . . And that's going to get you to contribute more content, and that's going to get you . . . more likes and comments. . . . It's a 9 social-validation feedback loop . . . exactly the kind of thing that a hacker like myself would come up with, because you're exploiting a 10 vulnerability in human psychology. . . . The inventors, creators — it's me, it's Mark [Zuckerberg], it's Kevin Systrom on Instagram, it's 11 all of these people — understood this consciously. And we did it anyway.313 12 13 Tellingly, many tech leaders, including individuals with inside knowledge of the defects of 14 Meta's social media products, either ban or severely limit their own children's access to screen 15 time and social media.314 Such leaders in the field include Tim Cook and former Facebook 16 executives Tim Kendall and Chamath Palihapitiya.315 17 18 19 20 313 Mike Allen, Sean Parker unloads on Facebook: "God only knows what it's doing to our children's brains," Axios (Nov. 9, 2017), https://www.axios.com/2017/12/15/sean-parker- 21 unloads-on-facebook-god-only-knows-what-its-doing-to-our-childrens-brains-1513306792. 314 Samuel Gibbs, Apple's Tim Cook: "I Don't Want My Nephew on a Social Network", The 22 Guardian (Jan. 19. 2018), https://www.theguardian.com/technology/2018/jan/19/tim-cook-i-dont- want-my-nephew-on-a-social- 23 network#:~:text=The%20head%20of%20Apple%2C%20Tim,it%20was%20announced%20on%2 0Friday; James Vincent, Former Facebook Exec Says Social Media is Ripping Apart Society, The 24 Verge (Dec. 11, 2017), https://www.theverge.com/2017/12/11/16761016/former-facebook-exec- ripping-apart-society. 25 315 Samuel Gibbs, Apple's Tim Cook: "I Don't Want My Nephew on a Social Network", The 26 Guardian (Jan. 19. 2018), https://www.theguardian.com/technology/2018/jan/19/tim-cook-i-dont- want-my-nephew-on-a-social- 27 network#:~:text=The%20head%20of%20Apple%2C%20Tim,it%20was%20announced%20on%2 0Friday; James Vincent, Former Facebook Exec Says Social Media is Ripping Apart Society, The 28 Verge (Dec. 11, 2017), https://www.theverge.com/2017/12/11/16761016/former-facebook-exec- ripping-apart-society. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 76 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 83 of 278 1 3. Meta intentionally designed product features to addict children and adolescents. 2 3 262. Meta designed Facebook and Instagram with harmful defects that users encounter 4 at every stage of interaction with the product. These defects, which have harmed Plaintiffs and 5 other adolescents that use the products, include but are not limited to: (a) recommendation 6 algorithms, fueled by extensive data collection, which are designed to promote use in quantities 7 and frequency harmful to adolescents; (b) product features that prey upon children's desire for 8 validation and need for social comparison; (c) product features that are designed to create harmful 9 loops of repetitive and excessive product usage; (d) lack of effective mechanisms, despite having 10 the ability to implement them, to restrict children's usage of the products; (d) inadequate parental 11 controls, and facilitation of unsupervised use of the products; and (e) intentionally placed obstacles 12 to discourage cessation of use of the products. 13 263. Facebook and Instagram have been designed, maintained, and constantly updated 14 by one of the world's most wealthy, powerful, and sophisticated corporations. Large teams of 15 expert data scientists, user experience ("UX") researchers, and similar professionals have spent 16 years fine-tuning these products to addict users. Every aspect of the products' interfaces, each layer 17 of their subsurface algorithms and systems, and each line of underlying code has been crafted by 18 brilliant minds. Every detail—the color of product icons, the placement of buttons within the 19 interface, the timing of notifications, etc.—is designed to increase the frequency and length of use 20 sessions. Therefore, it is impractical to create a comprehensive list of addictive, harm-causing 21 defects in the product until in-depth discovery occurs. Many product features, such as the inner 22 workings of Meta's algorithms, are secret and unobservable to users. Discovery during this 23 litigation will reveal additional detail about the defective, addictive, and harmful design of Meta's 24 products. 25 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 77 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 84 of 278 1 a. Facebook's and Instagram's algorithms maximize engagement, promoting use at levels and frequency that is harmful to kids. 2 3 264. Meta has invested its vast resources to intentionally design Facebook and Instagram 4 to be addictive to adolescents, all the while concealing these facts from its users, including 5 Plaintiffs, Consortium Plaintiffs, and the public. 6 265. In its original form, Meta's Facebook and Instagram algorithms ranked 7 chronologically, meaning that a particular user's feed was organized according to when content 8 was posted or sent by the people the user followed. In 2009, Meta did away with Facebook's 9 chronological feed in favor of engagement-based ranking; in 2016, it did the same on Instagram. 10 This "engagement-based" system meant that posts that received the most likes and comments were 11 highlighted first for users. But facing declining engagement, Meta redesigned its algorithms once 12 again in or around early 2018. This change prioritized "meaningful social interaction" ("MSI"), 13 with the goal of showing users content with which they were more likely to engage. The MSI- 14 oriented algorithms purportedly emphasize the interactions of users' connections; e.g., likes and 15 comments, and give greater significance to the interactions of connections that appear to be closest 16 to users. Meta's current algorithms consider a post's likes, shares, and comments, as well as a 17 respective user's past interactions with posts with similar characteristics, and displays the post in 18 the user's feed if it meets these and certain other benchmarks. 19 266. While Meta has publicly attempted to cast MSI as making time spent on its platforms 20 more "meaningful," MSI was actually just another way for Meta to increase user engagement on 21 Instagram and Facebook. While the feature increases the likelihood that an interaction will be 22 "meaningful" by Meta's definition—more likes, comments, and interactions—it does not consider 23 whether recommended content is "meaningful" to the user. This sets up users who may have reacted 24 negatively to upsetting or dangerous posts to see more of the same. That, in turn, can lead to what 25 Meta itself calls a "horrible feedback loop / downward spiral" 316—with negative reactions leading 26 the algorithm to present more posts that generate more negative reactions. 27 28 316 META3047MDL-003-00068860 at META3047MDL-003-00068861. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 78 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 85 of 278 1 267. In algorithmically generating users' feeds, Meta draws upon the vast amount of data 2 it collects about and from its users. Meta's algorithms combine the user's profile (e.g., the 3 information posted by the user on the product) and the user's dossier (the data collected and 4 synthesized by Meta, to which it assigns categorical designations) along with a dossier of similar 5 users.317 Meta's algorithms identify and rank recommended posts to optimize for various outcomes, 6 such as for time-spent by a user or for user engagement. More often than not, this has the effect that 7 Meta's algorithms direct users to alarming and aversive material. 318 8 268. Much of what Meta shows users is content that they did not sign up for. In a 2019 9 internal document, a Meta data scientist explained: "users have told us the pages they would like 10 to see content from, but we often override those explicit preferences because our predictions of 11 what will get shared and engaged with disagree." 319 This same employee pointed to additional data 12 demonstrating that users get relatively little connected content (content from pages they chose to 13 like) as opposed to unconnected content that is reshared by others, even as Meta knows that such 14 content is less valued by users. 320 15 269. Meta also optimizes the design of its products for overall "network value"—that is, 16 what will get the most downstream engagement by other users—rather than what that specific user 17 would like.321 As one Meta employee put it, "we show things to users that we think they have a 18 19 317 20 Instagram's former Head of Product Analytics defined "ranking" as "an ordering of content by importance or relevance" in a 2018 post titled "Is Ranking Good?". Haugen_00002372 at 21 Haugen_00002374. 318 Haugen_00006798 at Haugen_00006799 (observing that Meta's recommendation algorithms 22 "are prone to recommending harmful content."); Haugen_00024997 at Haugen_00024997 (conducting experiment showing that, in 3 weeks, "by following just … recommended content, 23 the test user's News Feed has become a near constant barrage of polarizing nationalist content, misinformation, and gore." (emphasis in original)); Haugen_00024997 at Haugen_00024998 24 ("when Watch isn't sure what you want, it seems to recommend a lot of softcore porn."); Haugen_00003739 at Haugen_00003740 ("[Instagram] is more 'successful' ranking harmful 25 content than benign content, and is more likely to mistakenly rank higher a harmful content than to mistakenly rank higher benign content."). 26 319 Haugen_00021247 at Haugen_000212448; see also Haugen_00006798 at Haugen_00006799 (Meta Research Scientist in 2019: "it's at best unclear whether users 'want' us to put unconnected 27 stories in their feed, even if they like some of them."). 320 Haugen_00021247 at Haugen_000212448. 28 321 Haugen_00021247 at Haugen_00021251. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 79 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 86 of 278 1 small chance of sharing, leading to comments between people who see it downstream over things 2 that have a greater chance of being explicitly liked by that user." 322 3 270. Through these algorithms, Meta intentionally supplants the content that users have 4 elected to see with content that it believes will drive more use and engagement. Thus, the products 5 that Meta touts as "[g]iv[ing] people the power to build community and bring[ing] the world closer 6 together," are actually designed in a way that prioritizes not social connection but product use at 7 all costs, even to the detriment of the health and safety of young people. 323 The result, for Meta, is 8 an increase in its bottom line. The result for young users is products that are so addictive that they 9 return again and again, even when the products push posts they're not interested in. 10 271. Meta knew that its engagement-based ranking algorithm (and its subsequent, 11 iterative MSI ranking algorithm) was structured in a way that mean that content which produces 12 intense reactions (i.e., strong engagement) triggers amplification by the apps. This propels users 13 into the most reactive experiences, favoring posts that generate engagement because they are 14 extreme in nature. Zuckerberg publicly recognized this in a 2018 post, in which he demonstrated 15 the correlation between engagement and sensational content that is so extreme that it impinges upon 16 Meta's own ethical limits, with the following chart: 324 While Zuckerberg went on to claim that 17 Meta had designed its algorithms to avoid this natural propensity of engagement-based algorithms, 18 his claim to the public is belied by the extensive internal and external research indicating Meta's 19 products did amplify extreme material. 20 21 22 23 24 25 26 322 Haugen_00021247 at Haugen_00021251. 27 323 Meta, Mission Statement, Meta, https://about.meta.com/company-info/. 324 28 Mark Zuckerberg, A Blueprint for Content Governance and Enforcement, Facebook, https://www.facebook.com/notes/751449002072082/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 80 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 87 of 278 1 2 3 4 5 6 7 8 9 10 11 272. Other documents show that Meta's employees also discussed their motive for 12 changing the design of the algorithm—namely, that users began to interact less with the product, 13 which became a worrisome trend for Meta's bottom-line. Meta's engagement-based algorithm 14 (including its MSI variant) exploited extreme content to drive more engagement, which, in turn, 15 helped Meta sell more of the digital ads that generated most of their revenue. In 2016, one Facebook 16 Tech Lead wrote: "[W]e only cared about things like time spent, open links, etc. That's what we 17 optimized for. That's what we used to define success and failure. And that's the problem." 325 18 273. Meta intentionally designed its MSI-focused algorithms to collect and analyze 19 several kinds of Plaintiffs' data: a user's profile, content the user reports, content the user posts, 20 content viewed, content engaged with, navigation paths, watch time, hover time (the amount of 21 time a user viewed a piece of content), whether a user mutes or unmutes a video, and whether a 22 user makes a full video screen, among other data. 326 Meta uses this data from adolescents to predict 23 what posts will capture the user's attention. Meta also tracks and utilizes data from various other 24 25 26 27 325 Haugen_00001033 at Haugen_00001033. 28 326 Haugen_00017177 at Haugen_00017177. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 81 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 88 of 278 1 sources, such as a user's off-product activities and the activities on websites that contain Facebook 2 or Instagram like or share buttons.327 3 274. Meta's algorithmic ranking is utilized in a variety of product features that are 4 designed by Meta to maximize user engagement. 5 275. For example, the Instagram product consists primarily of a never-ending and user- 6 specific Feed, which Instagram's data-driven algorithms generate for each user. In the app's 7 "Home" pane, this feed includes (but is not limited to) photos and videos posted by Instagram users 8 that the user has elected to "follow," as well as recommended photos and videos. In the app's 9 "Explore" pane, the feed consists almost exclusively of photos and videos from users the user has 10 not elected to "follow." In both cases, Instagram's algorithms evaluate each user's data to predict 11 what material will maximize their attention and time spent using the product, irrespective of what 12 the user wants to see. 13 276. Other "recommendation" features that are similarly algorithmically powered 14 include Facebook's Newsfeed, Instagram's Feed, Instagram Reels, Facebook Reels, Facebook 15 Watch (and its "For You" page), Accounts to Follow, People You May Know (introductions to 16 persons with common connections or backgrounds), Groups You Should Join, and Discover 17 (recommendations for Meta groups to join). 18 277. These product features work in combination to create and maintain a user's "flow- 19 state": a hyper-focused, hypnotic state, where bodily movements are reflexive and the user is totally 20 immersed in smoothly rotating through aspects of the social media product. 328 21 278. They also create a phenomenon that Meta employees refer to as "fee[d]ing the 22 spiral."329 Because of the structure of the defective algorithm powering features like Explore, 23 327 Allen St. John, How Facebook Tracks You, Even When You're Not on Facebook, Consumer 24 Reports (April 11, 2018), https://www.consumerreports.org/privacy/how-facebook-tracks-you- even-when-youre-not-on-facebook-a7977954071/. 25 328 See e.g., What Makes TikTok so Addictive?: An Analysis of the Mechanisms Underlying the World's Latest Social Media Craze, Brown Undergraduate J. of Pub. Health (2021), 26 https://sites.brown.edu/publichealthjournal/2021/12/13/tiktok/ (describing how IVR and infinite scrolling may induce a flow state in users). 27 329 META3047MDL-003-00121808 at META3047MDL-003-00121808. Meta employees 28 sometimes refer to this "spiral" as a "rabbit hole." See also META3047MDL-003-00077939 at META3047MDL-003-00077939. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 82 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 89 of 278 1 "someone feeling bad sees content that makes them feel bad, they engage with it, and then their IG 2 is flooded w[ith] it."330 Meta recognizes that Instagram users at risk of suicide or self-injury are 3 more likely to "encounter more harmful suicide and self-injury content (through explore, related, 4 follower suggestions, etc)."331 Because Meta's algorithm prioritizes engagement above all else, any 5 harmful feeling or impulse that users have are amplified by Instagram—which becomes an echo 6 chamber screaming their most upsetting thoughts back at them. 7 279. This phenomenon was cast into vivid relief when 14 year-old Molly Russell took 8 her own life after viewing reams of content related to suicide, self-injury, and depression on 9 Instagram and several other products. 332 During an official inquest investigating the role that social 10 media products played in her death, a Meta executive said that such content was "safe" for children 11 to see.333 The coroner rejected this claim, finding instead that Molly "died from an act of self-harm 12 whilst suffering from depression and the negative effects of on-line content" that she had not sought 13 out, but that the products' algorithms had pushed on her. 334 "The platform operated in such a way 14 using algorithms as to result, in some circumstances, of binge periods of images, video clips and 15 text some of which were selected and provided without Molly requesting them. These binge 16 periods … are likely to have had a negative effect on Molly…. In some cases, the content was 17 particularly graphic, tending to portray self-harm and suicide as an inevitable consequence of a 18 condition that could not be recovered from. The sites normalised her condition focusing on a limited 19 20 330 21 META3047MDL-003-00121808 at META3047MDL-003-00121808. 331 META3047MDL-003-00068863 at META3047MDL-003-00068905, META3047MDL-003- 22 00068878; see also META3047MDL-003-00042548 ("[P]eople who are suffering from depression and self-harm go down IG rabbit holes, and explore functionality compounds this 23 issue."). 332 Dan Milmo, Social Media Firms 'Monetising Misery', Says Molly Russell's Father After 24 Inquest, The Guardian (Sept. 20, 2022), https://www.theguardian.com/uk- news/2022/sep/30/molly-russell-died-while-suffering-negative-effects-of-online-content-rules- 25 coroner. 333 Ryan Merrifeld, Molly Russell Inquest: Instagram Boss Says Suicidal Posts Shouldn't Be 26 Banned From App, The Mirror (Sept. 26, 2022), https://www.mirror.co.uk/news/uk-news/molly- russell-inquest-instagram-boss-28085269. 27 334 Ryan Merrifeld, Molly Russell Inquest: Instagram Boss Says Suicidal Posts Shouldn't Be 28 Banned From App, The Mirror (Sept. 26, 2022), https://www.mirror.co.uk/news/uk-news/molly- russell-inquest-instagram-boss-28085269. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 83 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 90 of 278 1 and irrational view without any counterbalance of normality."335 The coroner further observed that 2 "[t]here was no age verification when signing up to the on-line platform" and that Molly's parents 3 "did not have access, to the material being viewed or any control over that material." 336 4 280. Disturbingly, years before Meta sent an executive to the inquest to tout its products 5 as "safe," Meta had conducted internal research which warned that there was a risk of "similar 6 incidents like Molly Russell" because algorithmic product features were "[l]eading users to 7 distressing content."337 8 281. Despite Molly's death, and notwithstanding Meta's research into dangerous 9 spirals—at one point dubbed the "Rabbithole project"—the company did nothing to stop harm to 10 its young users. Meta has been clear about the problem: for young users, "our recommendations 11 algorithms will start pushing you down a rabbit hole of more egregious content." 338 They have been 12 clear about potential solutions: targeted changes to the algorithm do lead to a "meaningful drop in 13 exposure" to problematic content.339 But they have been resistant to making changes, for the 14 explicit, profit-minded reason that such tweaks "came with a clear engagement cost." 340 15 b. Facebook's and Instagram's user interfaces are designed to create addictive engagement. 16 17 282. To further drive user engagement (and thereby drive data collection and advertising 18 revenue), Facebook and Instagram also utilize a series of design features that are carefully 19 calibrated to exploit users' neurobiology. These features work in tandem with algorithmic ranking 20 to promote addictive engagement. Meta understands this: "teens tell us that they try to take a break 21 335 Andrew Walker, H.M. Coroner, Regulation 28 Report to Prevent Future Deaths 2 (Oct. 13, 22 2022), https://www.judiciary.uk/wp-content/uploads/2022/10/Molly-Russell-Prevention-of- future-deaths-report-2022-0315_Published.pdf. 23 336 Andrew Walker, H.M. Coroner, Regulation 28 Report to Prevent Future Deaths 2 (Oct. 13, 24 2022), https://www.judiciary.uk/wp-content/uploads/2022/10/Molly-Russell-Prevention-of- future-deaths-report-2022-0315_Published.pdf. 337 25 META3047MDL-003-00043617 at META3047MDL-003-00043644. 338 META3047MDL-003-00077939; see also META3047MDL-003-00068860 at *60 (users 26 "seeking" bad experiences can "get into a rabbithole of getting more and more bad content on our surfaces."); META3047MDL-003-00087111 at 7112 (acknowledging that a majority of "negative 27 experiences" come from algorithmically-powered features like explore and hashtags). 339 META3047MDL-003-00077939. 28 340 META3047MDL-003-00077939. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 84 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 91 of 278 1 but feel compelled back onto the app."341 But it does not warn prospective or current users about 2 the following features or their safety risks, which are particularly harmful to Plaintiffs and other 3 adolescents. 4 283. First, Meta programs IVR into its products. Behavioral training via intermittent 5 rewards keeps users endlessly scrolling in search of a dopamine release, oftentimes despite their 6 desire to put their phone down and move onto other activities. Children, who are less likely to have 7 adequate impulse control than adults, are more susceptible to being drawn into this engineered 8 "flow state" and more likely to grow dependent on Facebook or Instagram. 9 284. Second, Facebook and Instagram utilize "Likes" to control the release of dopamine 10 in children. This feature, which Meta created for Facebook and "introduced … to the world" in 11 2010, allows users to indicate that they "Like" a post and visibly tallies the number of "Likes" any 12 given post has earned.342 Instagram launched in 2010 with the like feature built-in—a user can 13 "Like" a post simply by tapping a heart-shaped button. 14 285. As with a slot machine, users never know when a like will come. This conditions 15 them to stay on the app. But it also exacerbates issues of social comparison and feedback seeking, 16 creating detrimental effects on adolescent physical and mental health. Indeed, Meta knows from its 17 own internal research that the like feature negatively impacts its younger users. 343 In that research, 18 Meta acknowledged how much users care about the number of likes they received. 344 19 286. Despite this knowledge, Meta has expanded the likes feature in both Facebook and 20 Instagram. In December 2016, Meta began allowing users to like comments, not just posts. In 21 February 2022, Meta began allowing users to "Like" Instagram Stories. 345 Expanding the like 22 23 341 24 META3047MDL-003-00093303. 342 Ray C. He, Introducing new Like and Share Buttons, Meta (Nov. 6, 2013), 25 https://developers.facebook.com/blog/post/2013/11/06/introducing-new-like-and-share-buttons/. 343 See Haugen_00008207 at Haugen_00008210 (explaining the stress and anxiety that likes 26 cause teens). 344 Haugen_0008207 at Haugen_0008232. 27 345 Jhinuk Sen, Instagram is adding Likes to Stories so it doesn't clog up people's inboxes, 28 Business Today (Feb. 15, 2022), https://www.businesstoday.in/technology/news/story/instagram- is-adding-likes-to-stories-so-it-doesnt-clog-up-peoples-inboxes-322661-2022-02-15. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 85 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 92 of 278 1 feature has intensified and multiplied the body of feedback that teen users receive (or don't receive) 2 on their posts, preying on their desire to seek validation through comparison with others. 3 287. Meta's research confirms that hiding likes for all of its users would decrease social 4 comparison associated with like counts.346 Unfortunately, its research also demonstrated that hiding 5 likes would decrease the rates at which users click on advertisements (and thereby lower Meta's ad 6 revenue).347 7 288. For that reason—despite its ability to alleviate the negative impact of likes on 8 Plaintiffs and younger users—Meta chose only to implement half-measures. Meta created the 9 option for users to hide like counts in May 2021, but it made this an optional setting left off by 10 default.348 Moreover, the number of likes remain visible to the poster of the content. These changes 11 stop short of resolving the issue of negative social comparison that these score-keeping features 12 inflict. 13 289. Third, Meta has designed its video features in several ways geared to maximizing 14 users' flow state and keeping them immersed in its products for longer periods of time. Video clips 15 on Facebook Reels and Instagram Reels automatically play as a user scrolls and automatically 16 restart once they conclude. Reels cannot be paused and tapping on the video will simply mute its 17 audio. In addition, Meta imposes limits on the length of video content on Reels (currently 90 18 seconds, and at times as short as 15 seconds). These limits ensure that users do not become bored 19 by long videos and end their sessions. 20 290. Meta designed the comment features of Reels to minimize any disruption to users' 21 heightened flow state. The interface of Reels displays the "Like," "Comment," "Save," and "Share" 22 buttons on the bottom right of the screen. This placement avoids the milliseconds of delay or 23 discomfort that could disrupt the flow state of right-handed users if placed elsewhere on the screen. 24 Furthermore, these buttons are overlaid on top of the continuously playing clips, to eliminate any 25 temporal or visual interruption during which a user might evaluate whether to continue using the 26 346 Haugen_0008207 at Haugen_0008232. 27 347 Haugen_0008207 at Haugen_0008250. 348 28 Meta, Giving People More Control on Instagram and Facebook (May 26, 2021), https://about.fb.com/news/2021/05/giving-people-more-control/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 86 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 93 of 278 1 product. Likewise, when a user taps to view the comments on a Reel, the video's audio and the top 2 quarter of the video continue to play behind the comments section. Again, this design feature keeps 3 the user's attention on the feed. 4 291. In keeping with its study of IVR, Meta knows when to strategically interrupt a user's 5 flow. Occasionally, while a video is playing, a comment from the video will appear on the bottom 6 of the screen, even without the user tapping to view the comments section. These comments are 7 selected, displayed, and timed intentionally, to retain a user's attention by engaging with the 8 comments section. 9 292. Fourth, Meta carefully (and defectively) calibrates the notifications it sends outside 10 of the Facebook and Instagram apps, to maximize success in drawing back users who are not 11 presently using the products. By default, Facebook and Instagram notify users through text and 12 email about activity that might be of interest, which prompts users to open and reengage with the 13 products. However, Meta intentionally chooses to display only a limited amount of information in 14 notifications, in order to trigger curiosity and manipulate the user to click or tap through to the 15 product.349 In December 2020, Meta internally acknowledged that the goal of this feature was to 16 optimize engagement at the expense of value to users: "A few years ago we stopped sending out 17 emails telling you what happened - e.g., telling you what your friend did - instead we just say 18 'someone comment [sic] on your post,' in the hope that you'll click through. This a clear value- 19 engagement tradeoff."350 Similarly, Meta stopped sending push notifications about friend activities, 20 finding that without notifications, users were forced to go to the product itself to "check what's 21 happening," thereby initiating a new session, increasing engagement, and improving Meta's bottom 22 line.351 It designed these features despite knowledge that notifications were tied to potentially 23 addictive behaviors,352 and in disregard of safer alternative designs. 353 24 349 Clickbait, Merriam-Webster Dictionary, https://www.merriam- 25 webster.com/dictionary/clickbait. 350 Haugen_00010114 at Haugen_00010117. 26 351 Haugen_00010114 at Haugen_00010117. 352 See Haugen_00016893 at Haugen_00016899-902. 27 353 See Haugen_00016893 at Haugen_00016913 (suggesting changing to a subtler form of 28 notifications); Haugen_00016893 at Haugen_00016915- Haugen_00016916 (Meta intern urging PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 87 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 94 of 278 1 293. Meta's studied manipulation of user engagement through notifications is 2 particularly detrimental to teenagers, who lack impulse control and crave social rewards, and who 3 are therefore more susceptible to falling into compulsive patterns of product use. Those harms are 4 compounded by the fact that Meta sends push notifications in the middle of the night, prompting 5 children to re-engage with Instagram and Facebook the apps when they should be sleeping. 6 Disturbed and insufficient sleep is associated with poor health outcomes. 354 7 294. Fifth, the "Stories" feature of both Facebook and Instagram is defectively designed 8 to create artificial urgency so that users return to the apps. "Stories" was added by Meta in response 9 to the growing popularity of Snapchat with teenagers in 2016. "Stories" appear at the top of a user's 10 home page upon opening the app and are available to view for only 24 hours, after which they 11 disappear. This creates pressure to use the product daily, or else risk missing out on dopamine- 12 causing stimuli or social interactions. This feature is particularly addicting to adolescent users like 13 Plaintiffs, who feel increased social pressure to view all their contact's stories each day before the 14 content disappears, thus increasing their compulsive usage and potential addiction to the product. 355 15 The ephemeral nature of disappearing content is a ploy intended to inspire urgent perusal, and it 16 works.356 17 295. Sixth, Instagram and Facebook have recognized that their algorithms are structured 18 to recommend "keywords" or "hashtags" to its young users that lead them to navigate to dangerous 19 content.357 One researcher put the matter directly in April 2021: "A recurring area of concern is that 20 we are recommending keywords related to significant safety and wellbeing concerns e.g. weight 21 the company to stop "inundating users with excessive notifications" and instead focus on user 22 experience to create meaningful connections for users). 354 Nat'l Inst. of Mental Health, The Teen Brain: Still Under Construction 6 (2011), 23 http://www.ncdsv.org/images/NIMH_TeenBrainStillUnderConstruction_2011.pdf. 355 Sarah Lempa, Why Are Instagram Stories So Addicting?, Healthline (April 5, 2021), 24 https://www.healthline.com/health/why-are-instagram-stories-so-addicting#The-story-behind-the- Stories. 25 356 Madiha Jamal, Ephemeral Content — The Future of Social Media Marketing, Better Marketing (March 2, 2021), https://bettermarketing.pub/ephemeral-content-the-future-of-social- 26 media-marketing- 996d265916c2#:~:text=Ephemeral%20content%20relates%20to%20the,WhatsApp%20Stories%2 27 C%20and%20LinkedIn%20Stories. 357 28 See META3047MDL-003-00068863 at META3047MDL-003-00068905 ("We are leading users to content that can intensify their feelings through suggested/related hashtags"). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 88 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 95 of 278 1 loss, diet pills, appetite suppressants. We have been flagging these terms as they appear and Product 2 Policy and Product teams have been sweeping the list of keywords to remove them, but this is not 3 sustainable and remains a significant safety, policy, and comms risk. Our current approach of 4 catching all potentially risky terms in a 'block list' has not helped us avoid two news cycles, and 5 the possibility of this happening a third time is a significant comms and policy risk." 358 As another 6 set of Meta researchers acknowledged, the majority of negative experiences on Instagram come not 7 from direct interactions with others (i.e., through comments or direct messages) but rather through 8 algorithmically-generated recommendations, via Explore, Feed, or hashtags. 359 9 296. All of the above defects, in addition to the Instagram-specific defects discussed in 10 the section that follows, interact with and compound one another to make Meta's products addictive 11 and harmful for kids, including Plaintiffs. 12 297. Meta has long been aware of this dangerous and toxic brew. 13 298. In 2017, Meta investigated Facebook users who were addicted to the product—that 14 is, those who "cannot stop using [the] product to the point where it can cause them harm." 360 The 15 research found that, "[i]n a given week, approximately 5.9 million people leave Facebook" because 16 they "spent too much time" or because they were taking a temporary break and "planned to 17 return."361 "[T]his subset provided a good signal for people who could be addicted, who ultimately 18 leave Facebook as a solution.362 The analysis also found that this subset had a higher number of 19 sessions per day, received more notifications, and responded quicker to notifications compared to 20 all users.363 In 2018, Meta examined the issue of what its researchers called "Facebook addiction" 21 through a study titled "Problematic Facebook Use: When People Feel Like Facebook Negatively 22 Affects Their Life."364 The investigators defined "problematic use" as meaning: "Serious problems 23 24 358 META3047MDL-003-00184585 at META3047MDL-003-00184587. 359 META3047MDL-003-00087111 at 7112 25 360 Haugen_00016893 at Haugen_00016895. This group's investigation also included meeting 26 with Nir Eyal, author of the book Hooked: How to Build Habit-Forming Products. 361 Haugen_00016893 at Haugen_00016898. 27 362 Haugen_00016893 at Haugen_00016898. 363 Haugen_00016893 at Haugen_00016899-Haugen_00016802. 28 364 Haugen_00021690 at Haugen_00021690. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 89 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 96 of 278 1 with sleep, work or relationship that they attribute to Facebook AND concerns or preoccupations 2 about how they use Facebook (e.g., a fear of missing out (FOMO) or lack of control)." 365 Notably, 3 the investigators did not target the heaviest Facebook users in their research. 366 4 299. The study found that up to 5% of teens aged 13-20 were problematic users. 367 5 "Problematic use is highest among teens and people in their 20s, consistent with previous findings 6 that younger people generally have more problems with self-regulation. 368 Additionally, 7 "problematic users" evidenced common tendencies, such as (a) accessing and spending more time 8 on Facebook; (b) using Facebook late at night; (c) receiving more and responding more quickly to 9 push notifications; (d) temporarily deactivating their account in the past; and (e) sending far more 10 messages per minute with a higher ratio of messages sent to messages received. 369 As noted above, 11 Meta understands that "teens feel addicted to IG and feel a pressure to be present" but "like addicts, 12 they feel that they are unable to stop themselves from being on IG." 370 13 300. A study into Instagram user behaviors from that same year similarly found that "high 14 time spent users do tend to be disproportionately younger users, and these users may warrant extra 15 attention."371 The study found that "[a]s time spent increases, we see a larger proportion of users 16 that are high school, college or early work life-stages, with additional increases in high school when 17 we zoom in on the top 1% of time spent users." 372 18 301. Meta knows that "problematic use" of Facebook and Instagram leads to real 19 problems. In one internal company document, Meta acknowledged that the pressure to be present 20 and obtain validation on Instagram meant that teens lacked the capacity to "switch off and shut 21 22 23 365 Haugen_00021690 at Haugen_00021692. 24 366 Haugen_00021690 at Haugen_00021697. 367 25 Haugen_00021690 at Haugen_00021699. 368 Haugen_00021690 at Haugen_00021697. 26 369 Haugen_00021690 at Haugen_00021695-Haugen_00021697. 370 META3047MDL-003-00157036 at META3047MDL-003-00157036. 27 371 Haugen_00017177 at Haugen_00017181. 28 372 Haugen_00017177 at Haugen_00017187. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 90 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 97 of 278 1 down," noting that teens "can get addicted to things that make them feel bad." 373 One of Meta's 2 data scientists did not mince words when describing this phenomenon to their colleagues: 3 I worry that driving sessions incentivize us to make our product more addictive, without providing much more value. How to keep 4 someone returning over and over to the same behavior each day? Intermittent rewards are most effective (think slot machines), 5 reinforcing behaviors that become especially hard to distinguish— even when they provide little reward, or cease providing reward at 6 all.374 7 Another Meta employee was clear-eyed that "little reward" was too charitable—and that 8 addictive use was actively harming kids' mental health: 9 In the focus groups teens told us that they don't like the amount of time they spend on the app but feel like they have to be present. They 10 often feel 'addicted' and know that what they're seeing is bad for their mental health but feel unable to stop themselves. This makes 11 them not feel like they get a break [sic] or to can't switch off social media. . . . 12 [A]bout 30% (and an even larger proportions of those who are 13 unsatisfied with their lives) said that the amount of time they spend on social media makes them feel worse. About half of teens in both 14 markets want Instagram to take a break or to get off the app. . . . [In another survey], we found that time spent is among one of the most 15 negative experiences for IG (25%+ say they spend too much time on social media and it's worst on Instagram and Facebook). At the same 16 time, they didn't think there was anything they could do about it and had fairly negative things to say about the time spent tools we have 17 (particularly that the tools are easy to ignore). 375 18 302. In January 2021, another Meta employee wrote: "No one wakes up thinking they 19 want to maximize the number of times they open Instagram that day. But that's exactly what our 20 product teams are trying to do."376 21 303. Meta failed to invest in adequate tools to limit the harm their products inflicted on 22 users. As one employee candid put it: "the tools we currently have aren't effective at limiting 23 [users'] time on the app."377 Nonetheless, Meta publicly presented certain of these tools as 24 25 373 Haugen_00017069 at Haugen_00017128, Haugen_00017132. 26 374 Haugen_00010114 at Haugen_00010127. 27 375 Haugen_00017069 at Haugen_00017171. 376 META3047MDL-003-00161686 at META3047MDL-003-00161686. 28 377 META3047MDL-003-00157036 at META3047MDL-003-00157036. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 91 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 98 of 278 1 solutions, despite knowing of their ineffectiveness. For example, Meta offered to its users a feature 2 that purported to show how much time users had spent on Instagram. And Meta touted this feature 3 "when speaking to consumers, the press, and stakeholders about our efforts to combat social media 4 addiction."378 But internally, Meta acknowledged that the data reported by this tool was 5 fundamentally "incorrect": "It's not just that Apple / Google have better data. Ours is wrong. Far 6 worse. We're sharing bad metrics externally. We've been unable to right it despite several person- 7 months of efforts. . . . So it's wrong (bad enough in itself), can't be fixed easily (we've tried), has 8 been half-rolled-out for a while . . . the group that audits metrics we provide to the outside world, 9 has called us out on it…The reason this is relevant is we vouch for these numbers. Any day they're 10 out there is a legal liability."379 11 304. At the end of the day, Meta's failure to prevent compulsive use by children, and the 12 harms resulting therefrom, are a simple function of its misplaced priorities. One "integrity 13 researcher" at Facebook wrote an internal article in August 2020 with her parting thoughts as she 14 left the company. She explained that Meta's leadership consistently ignored concerns about user 15 safety: 16 Integrity teams are facing increasing barriers to building safeguards. . . . [T]ime and time again I've seen promising interventions from 17 integrity product teams, with strong research and data support be prematurely stifled or severely constrained by key decision makers— 18 often based on fears of public and policy stakeholder responses. Similarly (though even more concerning), I've seen already built & 19 functioning safeguards being rolled back for the same reasons . . . While mountains of evidence is (rightly) required to support a new 20 intervention, none is required to kill (or severely limit) one. . . . [This] is intended as a call to reflection for those decision-makers imposing 21 constraints.380 22 305. Meta's decision to hook teenage users by rewiring their brains has not aged well for 23 some of its former employees. Chamath Palihapitiya, the former Vice President of User Growth at 24 Facebook, admitted that he feels "tremendous guilt" about his contributions to social media, saying 25 26 27 378 META3047MDL-003-00157036 at META3047MDL-003-00157036. 379 META3047MDL-003-00157133 at META3047MDL-003-00157133. 28 380 Haugen_00021096 at Haugen_00021097-Haugen_0002110 (emphasis omitted). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 92 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 99 of 278 1 "[t]he short-term, dopamine-driven feedback loops that we have created are destroying how society 2 works."381 3 c. Instagram's defective product features cause negative appearance comparison and social comparison 4 5 306. Meta has known since at least 2018 that Instagram has a corrosive effect on the well- 6 being of pre-teen and teenage users. 382 7 307. Meta has an internal research team comprised of employees with expertise in, inter 8 alia, computer science, psychology, and quantitative and qualitative analysis. From 2019 to 2021, 9 this team conducted a "teen mental health deep dive" which included focus groups, diary studies, 10 and online surveys. One large-scale study paired a survey of tens of thousands of Instagram users 11 with data about the time each respondent spent on Instagram and the type of content they viewed. 383 12 308. The evidence collected by Meta's research team is damning. Among other findings, 13 Defendants' researchers learned that: 14  41% of teen users of Instagram in the U.S. and U.K. who reported feeling 15 "unattractive" said the feeling began while using the product; 384 16 17 18 19 381 Amy B. Wang, Former Facebook VP says social media is destroying society with 'dopamine- 20 driven feedback loops', Wash. Post (Dec. 12, 2017), https://www.washingtonpost.com/news/the- switch/wp/2017/12/12/former-facebook-vp-says-social-media-is-destroying-society-with- 21 dopamine-driven-feedback-loops/. 382 See, e.g., Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is 22 Toxic for Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- 23 documents-show-11631620739; META3047MDL-003-00146240 at META3047MDL-003- 00146256. 24 383 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), 25 https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- documents-show-11631620739; Haugen_00017069; META3047MDL-003-00000029. 26 384 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for 27 Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- 28 documents-show-11631620739; META3047MDL-003-00000029 at META3047MDL-003- 00000043. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 93 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 100 of 278 1  32% of teenage girls said that when they felt bad about their bodies, 2 Instagram made them feel worse;385 3  "We make body issues worse for 1 in 3 teen girls;"386 4  "Frequent social comparison is a key driver of subjective well-being and 5 teens say IG makes this problem worse;"387 6  One in five teens say that Instagram makes them feel worse about 7 themselves;388 8  Two-thirds of teen girls on Instagram experience negative social 9 comparison;389 10  17% of teen girl Instagram users say the product makes "[e]ating [i]ssues" 11 worse;390 12  About a quarter of teens who reported feeling "not good enough" said the 13 feeling started on Instagram;391 14  Many teens said Instagram undermined their confidence in the strength of 15 their friendships;392 16 385 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for 17 Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- 18 documents-show-11631620739; Haugen_00019219 at Haugen_00019226; META3047MDL- 003-00001846 at META3047MDL-003-00001852. 19 386 Haugen_00016699 at Haugen_00016707. 387 Haugen_00019219 at Haugen_00019226. 20 388 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for 21 Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- 22 documents-show-11631620739; Haugen_00017069 at Haugen_00017091; META3047MDL- 003-00000029 at META3047MDL-003-00000049. 389 23 Haugen_00019219 at Haugen_00019226. 390 Haugen_00020135 at Haugen_00020162. 24 391 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), 25 https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- documents-show-11631620739; META3047MDL-003-00000029 at META3047MDL-003- 26 00000043. 392 27 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), 28 https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 94 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 101 of 278 1  Teenagers who struggle with mental health say that Instagram worsens those 2 problems;393 3  "Teens blame Instagram for increases in the rates of anxiety and depression 4 among teens" in recent years—a response that was unprompted and 5 consistent across all groups;394 6  Among teens who reported suicidal thoughts, 13% of British users and 6% 7 of American users traced the desire to kill themselves to Instagram; 395 and 8  13.5% of teen girl Instagram users say the product makes thoughts of 9 "suicide and self-injury" worse. 396 10 309. Meta's researchers were clear in explaining that Instagram product features were 11 responsible for these problems. In one chart illustrating the "High" amount of "Body, Appearance 12 Comparison" on Instagram, researchers cited as contributing factors "Product mechanics 13 (addicting)" and "Explore, discover, stalk (down the rabbit hole)." 397 In another slide, researchers 14 noted the particular problems with Instagram's Explore feature, as it contains "[t]ons of body image 15 triggers" that are "[i]Intimidating" to users.398 16 17 18 documents-show-11631620739 META3047MDL-003-00000029 at META3047MDL-003- 00000043. 19 393 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), 20 https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- documents-show-11631620739; META3047MDL-003-00000029 at META3047MDL-003- 21 00000054. 394 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for 22 Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- 23 documents-show-11631620739; META3047MDL-003-00000029 at META3047MDL-003- 00000052. 24 395 Georgia Wells, Jeff Horwitz, Deepa Seetharaman, Facebook Knows Instagram is Toxic for 25 Teen Girls, Company Documents Show, Wall St. J. (Sept. 14, 2021), https://www.wsj.com/articles/facebook-knows-instagram-is-toxic-for-teen-girls-company- 26 documents-show-11631620739; META3047MDL-003-00000029 at META3047MDL-003- 00000043. 396 27 Haugen_00016699 at Haugen_00016707. 397 Haugen_00015958 at Haugen_00015987. 28 398 Haugen_00015958 at Haugen_00015989. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 95 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 102 of 278 1 310. Children are developmentally unprepared for the psychological ramifications of 2 peer judgment and online comparisons. 3 311. Meta's internal researchers were not only clear about the fact that Instagram causes 4 a high level of social comparison for teenagers; they were clear-eyed about the dire consequences. 5 They observed that the addictive nature of the Instagram product, combined with a tendency for 6 users to share only the best moments and a pressure to match unrealistic beauty ideals can send 7 teens into a downward spiral that includes anger, withdrawal, insecurity, and body dysmorphia— 8 "a series of emotions that in many ways mimic stages of grief." 399 They further warned that 9 "[u]sers['] experience of [this] downward spiral is exacerbated by our platform." 400 "Comparisons 10 on Instagram can change how young women view and describe themselves," they noted, changing 11 a girl's self-perception from "multi-dimensional" and "centered" to "not in control," "dark," boxed 12 in," "low esteem," and "anxious." 401 The researchers' conclusions were stark: "Mental health 13 outcomes related to this can be severe," and can include "eating disorders," "body dysmorphia," 14 "body dissatisfaction," "depression," and "loneliness." 402 15 312. Meta's research demonstrates that social comparison is particularly bad on 16 Instagram because, among other things, celebrity and influencer content is pervasive. 403 By 17 manufacturing and emphasizing influence and celebrity, and purposely inundating tween and teen 18 users with those accounts, Meta further exploits and monetizes social comparison. That has come 19 at a direct cost to the mental health of its teens users, who are more susceptible to body 20 dissatisfaction and negative social comparisons.404 Meta knows as much. In 2021, its researchers 21 found that exposure to content from "Top Accounts" (i.e., those with the top 0.1% of followers) 22 was most associated with negative comparison and that Instagram's influence-driven model ensures 23 Top Accounts flood users' feeds almost half the time. 405 24 399 Haugen_00015958 at Haugen_00015985. 400 25 Haugen_00015958 at Haugen_00015990. 401 Haugen_00015958 at Haugen_00015983. 26 402 Haugen_00015958 at Haugen_00015992. 403 27 Haugen_00015958 at Haugen_00015996. 404 Haugen_00002527 at Haugen_00002555. 28 405 META3047MDL-003-00159559 at META3047MDL-003-00159560. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 96 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 103 of 278 1 313. Score-keeping features designed into Instagram amplify these problems. Teenage 2 girls are particularly impacted when comparing like counts, follower counts, views, and comments 3 on their posts to those of models, celebrities, and so-called influencers. Meta's internal research 4 reveals that teen girls are eight times more likely to engage in negative social comparison than their 5 male counterparts.406 6 314. Instagram compounds the foregoing problems with yet another pernicious feature— 7 image "filters" that allow users to engage in selective self-presentation by altering their appearance 8 in photos and videos. These filters allow facial structure alteration, body slimming, skin lightening, 9 skin tanning, blemish clearing, the artificial overlap and augmentation of makeup, and other 10 beautification "improvements."407 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 406 Haugen_00017263 at Haugen_00017263. 27 407 T. Mustafa, An 'Instagram Vs Reality' filter is showing how toxic photo editing can be, Metro 28 (April 2021); https://metro.co.uk/2021/04/30/an-instagram-vs-reality-tool-is-showing-how-toxic- filters-can-be-14498265/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 97 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 104 of 278 408 1 These filters have harmed Plaintiffs in multiple ways, both independently and in concert with 2 Instagram's other defective features.409 3 315. First, the easy accessibility of filters, combined with features such as "Likes," 4 encourage adolescents to artificially change their appearances. 410 As noted, adolescents naturally 5 seek social validation. When they notice increased interaction and favorable responses to their 6 filter-edited photos (more "Likes" and comments"), many are led to believe they are only attractive 7 when their images are edited.411 These young people, including Plaintiffs, begin to prefer how they 8 look using filters, not as they appear naturally.412 In a 2016 study, 52% of girls said they use image 9 filters every day, and 80% have used an app to change their appearance before age 13. 413 Meta's 10 own findings showed teen girls spend hours editing images by altering their appearance before 11 posting on Instagram,414 and that "teen girls in particular" are "some of the biggest users of these 12 filters."415 Pictures must be "Instagrammable" to be worthy of posting. 13 316. Second, because Instagram already promotes a high degree of social comparison, 14 youth including Plaintiffs find themselves comparing their real-life appearances to the edited 15 16 408 Mariska Kleemans, Serena Daalmans, Ilana Carbaat & Doeschka Anschütz (2018) Picture 17 Perfect: The Direct Effect of Manipulated Instagram Photos on Body Image in Adolescent Girls, 21 Media Psychology 93, 93-110 (2018), 18 https://www.tandfonline.com/doi/pdf/10.1080/15213269.2016.1257392. 409 Anna Haines, From 'Instagram Face' To 'Snapchat Dysmorphia': How Beauty Filters Are 19 Changing The Way We See Ourselves, Forbes (Apr. 27, 2021), https://www.forbes.com/sites/annahaines/2021/04/27/from-instagram-face-to-snapchat- 20 dysmorphia-how-beauty-filters-are-changing-the-way-we-see-ourselves/?sh=3c32eb144eff. 410 Tate Ryan-Mosley, Beauty Filters Are Changing the Way Young Girls See Themselves, MIT 21 Tech. Rev. (Apr. 2, 2021), https://www.technologyreview.com/2021/04/02/1021635/beauty- filters-young-girls-augmented-reality-social-media/amp/. 22 411 Tate Ryan-Mosley, Beauty Filters Are Changing the Way Young Girls See Themselves, MIT Tech. Rev. (Apr. 2, 2021), https://www.technologyreview.com/2021/04/02/1021635/beauty- 23 filters-young-girls-augmented-reality-social-media/amp/. 412 24 Poojah Shah, How Social Media Filters Are Affecting Youth, Parents (Apr. 28, 2022), https://www.parents.com/kids/health/childrens-mental-health/how-social-media-filters-are- 25 affecting-youth/. 413 Anna Haines, From 'Instagram Face' to 'Snapchat Dysmoprhia'; How Beauty Filters Are 26 Changing the Way We See Ourselves, Forbes (Apr. 27, 2021), https://www.forbes.com/sites/annahaines/2021/04/27/from-instagram-face-to-snapchat- 27 dysmorphia-how-beauty-filters-are-changing-the-way-we-see-ourselves/?sh=3c2d58704eff. 414 Haugen_00019219 at Haugen_00019255. 28 415 META3047MDL-003-00157020 at META3047MDL-003-00157020. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 98 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 105 of 278 1 appearances not only of themselves but of others online. 416 These false and unrealistic body image 2 standards further lead teenagers, including Plaintiffs, to develop negative perceptions of their 3 appearance. 77% of girls reported trying to change or hide at least one part of their body before 4 posting a photo of themselves, and 50% believe they did not look good without editing. 417 5 317. Third, the specific changes filters make to an individual's appearance can cause 6 negative obsession or self-hatred surrounding aspects of their appearance. 418 The filters alter 7 specific facial features such as eyes, lips, jaw, face shape, and slimness, which often require medical 8 intervention to alter in real life.419 The pervasiveness of Meta-designed filters through the algorithm 9 permeates Instagram and causes adolescent users to negatively compare their real appearances 10 against a false physical reality.420 In one recent study, even users who reported a higher initial self- 11 esteem level felt they looked 44% worse before their image was edited using a filter. 421 "[W]hen 12 the . . . filter increased the gap between how participants wanted to look and how they felt they 13 actually looked, it reduced their self-compassion and tolerance for their own physical flaws." 422 As 14 15 16 416 See Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the U.S., Wall. St. J. (Sept. 29, 2021), https://s.wsj.net/public/resources/documents/teen-girls- 17 body-image-and-social-comparison-on-instagram.pdf (explaining that users forget that Instagram only shows the highlights of people's lives and is not depicting reality); Haugen_00019219 at 18 Haugen_00019255. 417 Anna Haines, From 'Instagram Face' to 'Snapchat Dysmorphia'; How Beauty Filters Are 19 Changing the Way We See Ourselves, Forbes (Apr. 27, 2021), https://www.forbes.com/sites/annahaines/2021/04/27/from-instagram-face-to-snapchat- 20 dysmorphia-how-beauty-filters-are-changing-the-way-we-see-ourselves/?sh=3c2d58704eff. 418 21 Tonya Russell, Social Media Filters Are Changing How Young People See Themselves, Teen Vogue (Jan. 25, 2022), https://www.teenvogue.com/story/social-media-filters-how-young-people- 22 see-themselves/amp. 419 Tonya Russell, Social Media Filters Are Changing How Young People See Themselves, Teen 23 Vogue (Jan. 25, 2022), https://www.teenvogue.com/story/social-media-filters-how-young-people- see-themselves/amp. 24 420 Tonya Russell, Social Media Filters Are Changing How Young People See Themselves, Teen Vogue (Jan. 25, 2022), https://www.teenvogue.com/story/social-media-filters-how-young-people- 25 see-themselves/amp. 421 Ana Javornik, Ben Marder, Marta Pizzetti, & Luk Warlop, Research: How AR Filters Impact 26 People's Self-Image, Harvard Business Review (December 22, 2021), https://hbr.org/2021/12/research-how-ar-filters-impact-peoples-self-image. 27 422 Ana Javornik, Ben Marder, Marta Pizzetti, & Luk Warlop, Research: How AR Filters Impact 28 People's Self-Image, Harvard Business Review (December 22, 2021), https://hbr.org/2021/12/research-how-ar-filters-impact-peoples-self-image. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 99 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 106 of 278 1 one psychodermatologist has summed it up, "these apps subconsciously implant the notion of 2 imperfection and ugliness, generating a loss of confidence." 423 3 318. Fourth, Meta has intentionally designed its product to not alert adolescent users 4 when images have been altered through filters or edited. Meta has therefore designed its product so 5 that users, including Plaintiffs, cannot know which images are real and which are fake, deepening 6 negative appearance comparison. 7 319. The impact of the negative social and appearance comparison caused by Meta's 8 defective product features is profound. Instagram-induced social comparison creates a schism 9 between the ideal self and the real self, leading to distress and depression. Filters, especially in 10 combination with other product features, cause body image issues, eating disorders, body 11 dysmorphia, and related harms.424 12 320. Again, none of this has been lost on Meta. To the contrary, Meta has long been 13 aware of the harms Instagram inflicts on youth by perpetuating social comparison to unrealistic 14 beauty standards. In one study from 2019, teens age 13-17 explained that Instagram harms their 15 mental health by creating pressure to conform to social stereotypes and match the body shapes of 16 influencers, the need for validation through views, likes and followers, and the over-sexualization 17 of girls.425 Meta's analysis categorized the documented harms into three categories: impacts from 18 comparison to others, the pressure of looks/behaviors, and from others' behaviors. These impacts 19 20 21 423 Genesis Rivas, The Mental Health Impacts of Beauty Filters on Social Media Shouldn't Be 22 Ignored – Here's Why, InStyle (Sept. 14, 2022), https://www.instyle.com/beauty/social-media- filters-mental-health. 23 424 See Sian McLean, Susan Paxton, Eleanor Wertheim, & Jennifer Masters, Photoshopping the Selfie: Self Photo Editing and Photo Investment Are Associated with Body Dissatisfaction in 24 Adolescent Girls, 48 Int'l J. of Eating Disorders 1132, 1133 (Aug. 27, 2015), https://pubmed.ncbi.nlm.nih.gov/26311205/ (presenting a 2015 study involving 101 adolescent 25 girls, which found that more time spent editing and sharing selfies on social media raised their risk of experiencing body dissatisfaction and disordered eating habits.); Scott Griffiths, Stuart 26 Murray, Isabel Krug, & Sian McLean, The Contribution of Social Media to Body Dissatisfaction, Eating Disorder Symptoms, and Anabolic Steroid Use Among Sexual Minority Men, 21 27 Cyberpsychology Behavior, and Soc. Networking 149, 149 (Mar. 1, 2018), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5865626/. 28 425 Haugen_00017069 at Haugen_00017122. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 100 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 107 of 278 1 were associated with isolation, unhealthy eating habits, depression, anxiety, insecurity, and 2 loneliness.426 3 321. In its "Social Comparison Exploratory Research" conducted in 2020, Meta 4 acknowledged that body image comparisons are formed in part by its defective product features— 5 filters that flood its app with seemingly unattainable looks like flawless skin, made worse by posters 6 "using hashtags like no-filters but actually using filters."427 Meta's researchers found that nearly 7 half of teen girls on Instagram feel like they often or always compare their appearance to others 8 using the product, more than one-third feel extreme pressure to look perfect on Instagram. 428 In a 9 related survey, Meta found that around the age of 30, the role of Instagram in social comparison 10 begins to diminish.429 11 322. According to research conducted by Meta in 2019, over 60% of teens believe 12 Instagram should help them address the effects of social comparison by recommending positive 13 accounts, reprioritizing their feeds to promote healthy accounts, and help them follow a balance of 14 accounts.430 One in three teens wished Instagram gave them better user controls. 431 Yet a survey 15 conducted two years later revealed that Meta had done little to address its users' concerns. Topics 16 that elicited social comparison still encompassed over one-third of teen girls' feeds. And for every 17 post from a friend that appeared in a teen girl's feed, Instagram's algorithm drove five times as 18 much content from popular accounts. 432 19 323. One slide from Meta's study of social comparison offers a particularly succinct 20 summation of how the various product defects built into Instagram "exacerbate each other to create 21 22 23 426 Haugen_00017069 at Haugen_00017126. 427 Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the 24 U.S., Wall. St. J. (Sept. 29, 2021), https://s.wsj.net/public/resources/documents/teen-girls-body- image-and-social-comparison-on-instagram.pdf; Haugen_00015958 at Haugen_00015971- 25 Haugen_00015977. 428 26 Haugen_00007080 at Haugen_00007082. 429 Haugen_00007080 at Haugen_00007095. 27 430 Haugen_00017069 at Haugen_00017145. 431 Haugen_00020135 at Haugen_00020171. 28 432 Haugen_00002527 at Haugen_00002527. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 101 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 108 of 278 1 a perfect storm."433 "Posting 'For the Gram'" creates a "Pressure to Look Perfect." 434 The ability 2 of influencers to "Monetiz[e] face + body" creates a "Highlight Reel Norm." 435 And the "Vortex 3 of Feed + Profile and Explore" promotes a "Hate to love" dynamic for users, which "Feed[s] the 4 Spiral" of compulsive use.436 Taken together, these three features—all driven by very specific 5 design features of Instagram—create a "Social Comparison Sweet Spot." 437 6 7 8 9 10 11 12 13 14 15 16 17 324. Finally, Meta understands that the social comparison it knowingly enables through 18 appearance filters create compulsive behavior among child users, especially when paired with other 19 20 433 Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the 21 U.S., Wall. St. J. (Sept. 29, 2021), https://s.wsj.net/public/resources/documents/teen-girls-body- image-and-social-comparison-on-instagram.pdf; Haugen_00015958 at Haugen_00015991. 22 434 Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the 23 U.S., Wall. St. J. (Sept. 29, 2021), https://s.wsj.net/public/resources/documents/teen-girls-body- image-and-social-comparison-on-instagram.pdf; Haugen_00015958 at Haugen_00015991. 435 24 Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the U.S., Wall. St. J. (Sept. 29, 2021), https://s.wsj.net/public/resources/documents/teen-girls-body- 25 image-and-social-comparison-on-instagram.pdf; Haugen_00015958 at Haugen_00015991. 436 Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the 26 U.S., Wall. St. J. (Sept. 29, 2021), https://s.wsj.net/public/resources/documents/teen-girls-body- image-and-social-comparison-on-instagram.pdf; Haugen_00015958 at Haugen_00015991. 27 437 Teen Girls Body Image and Social Comparison on Instagram – An Exploratory Study in the 28 U.S., Wall. St. J. (Sept. 29, 2021), https://s.wsj.net/public/resources/documents/teen-girls-body- image-and-social-comparison-on-instagram.pdf; Haugen_00015958 at Haugen_00015991. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 102 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 109 of 278 1 defects such as likes and algorithmic recommendations. Specifically, Meta knows that social 2 comparison creates a negative feedback loop. 438 Its internal research reveals that, as teens compare 3 themselves to others, their self-doubt grows, which in turn heightens the degree of attention they 4 give these feelings. As these effects compound, teens experience depression and anxiety, making 5 them more vulnerable and susceptible to harmful content. 439 Meta observed that long-term constant 6 self-critique and scrutiny permanently shapes how teens view themselves in all relationships on 7 and offline.440 Moreover, they found that the incessant social pressure Instagram inflicted on teens 8 led to obsessive control and attention-seeking behavior to obtain social validation. 441 In other 9 words, Instagram's design features resulted in an insidious cycle of harm where teens believed they 10 could only find reprieve by increased Instagram use. 11 325. Meta has the technological capabilities to mitigate social comparison harms 12 significantly but actively chooses to ignore leading research (including its own) and its product 13 engineers' recommendations. One internal presentation concluded with several "targeted 14 interventions" for changes to Instagram that could mitigate these harms, such as a recommendation 15 that users take a break during a long use session. 442 In another, computational social researchers 16 and engineers at Meta proposed numerous feasible product design changes including: demotions 17 on Explore and Reels using topic and image and video features from an FBLearner model, 18 separating top-account feed from close-friend feed, and not recommending celebrities to follow 19 that post primarily fashion/beauty content as users "can find these accounts on their own, but [Meta] 20 shouldn't amplify their influence through recommendations." 443 21 326. Despite its vast knowledge of the harms that Instagram's defective product features 22 were causing to adolescents, in Meta's 2021 Milestone Tracker, the action item of reducing the 23 negative effects from social comparison through controls had yet to be started. 444 In other words, 24 438 Haugen_00017069 at Haugen_00017127. 25 439 Haugen_00017069 at Haugen_00017127. 440 26 Haugen_00017069 at Haugen_00017130. 441 Haugen_00017069 at Haugen_00017127. 27 442 Haugen_00019219 at Haugen_00019272. 443 Haugen_00002527 at Haugen_00002565. 28 444 Haugen_00025741 at Haugen_00025763. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 103 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 110 of 278 1 despite awareness that the deliberate design of Instagram was drastically damaging teen mental and 2 physical health, including that of Plaintiffs, Meta ignored the problem, failing to implement its own 3 researchers' recommendations. 4 d. Meta has failed to implement effective age-verification measures to keep children off of Facebook and Instagram. 5 6 327. Meta purports to ban children under the age of 13 from using its products but, at all 7 relevant times, has lacked any reliable form of age verification to prevent such use. 8 328. Other online products employ substantially more effective and reliable age 9 verification schemes before granting children access. These include but are not limited to 10 connecting new users to parents' accounts, credit card verification, verification by presentation of 11 identification card (or other government-issued document), or linking a verified undergraduate or 12 professional email, among other methods. Meta chooses not to implement any of these systems, 13 even though they are technologically feasible, used by many companies across the Internet, and 14 could be employed at relatively low cost. Indeed, Meta itself uses an age verification technique for 15 its Facebook Dating product that it claims can verify ages without identifying users—but does not 16 use the same technology at account startup for Facebook or Instagram. 445 17 329. For most of its history, Meta knew that children under the age of 13 were using its 18 apps. And it certainly could have figured this out based on posted photos of elementary school age 19 users. Yet Meta continued to promote and target Facebook and Instagram to children. As long as a 20 new user simply clicked a box confirming that they were at least 13 years old, Meta asked no 21 questions, engaged in zero follow-up, and let the user access the products indefinitely. This did not 22 go unnoticed by certain of its employees, who criticized the company's policy: "if we collected age 23 on IG we could age-gate this content [referring to suicide and self-injury ("SSI") content] . . . and 24 if we used age classifiers we could detect under 13s and kick them off the platform so they wouldn't 25 have access to content that's not appropriate for them to find." 446 26 445 Erica Finkle, Meta Director of Data Governance, Bringing Age Verification to Facebook 27 Dating, Meta (Dec. 5, 2022), https://about.fb.com/news/2022/12/facebook-dating-age- 28 verification/. 446 META3047MDL-003-00086015 at META3047MDL-003-00086015. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 104 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 111 of 278 1 330. Indeed, Meta did not ask for the age of new Instagram users until December 2019, 2 after Instagram had been on the market for more than seven years. 447 Even then, Meta did not ask 3 existing users to disclose their ages, effectively grandfathering in underage users. Indeed, an 4 internal document confirms that, in April 2020, Meta only had ages for approximately 55% of its 5 users.448 Meta did not attempt to correct that problem until August 30, 2021. And Meta did not 6 begin requiring age verification for users who attempt to change their age from under to over 18 7 until 2022.449 8 331. But even this minimal age verification procedure is toothless. Well over a year since 9 the Congressional testimony of whistleblower Frances Haugen regarding the harm Meta causes to 10 minors, Meta still does not as a default matter require users to verify their ages upon signing up to 11 use Instagram or Facebook. Users are only asked to self-report their birthday: 12 13 14 15 16 17 18 19 20 21 22 447 META3047MDL-003-00157020 at META3047MDL-003-00157020 ("[W]e have very limited 23 age information on IG (we only started collecting age in December at registration)"). 448 META3047MDL-003-00042548 at META3047MDL-003-00042551- META3047MDL-003- 24 00042552. 449 25 Instagram, Introducing New Ways to Verify Age on Instagram, Meta (June 23, 2022), https://about.fb.com/news/2022/06/new-ways-to-verify-age-on-instagram/. Meta explained the 26 choice of age by saying that they provide users under 18 with an experience that is appropriate for their age, including "preventing unwanted contact from adults they don't know." Instagram, 27 Introducing New Ways to Verify Age on Instagram, Meta (June 23, 2022), https://about.fb.com/news/2022/06/new-ways-to-verify-age-on-instagram/. However, as described 28 below, each week hundreds of thousands of children are inappropriately contacted by adults on Instagram. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 105 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 112 of 278 1 2 3 4 5 6 7 8 9 10 11 12 13 332. If the user reports a birthday indicating they are less than 13 years old, they are 14 informed that they cannot create an account. However, after acknowledging this message, users can 15 immediately reattempt to create an account and input an eligible birthday. When a user enters an 16 eligible birthday, there are no restrictions to creating an account, other than having it linked to a 17 cell phone number or an email address. In other words, Meta routinely allows pre-teens to 18 misrepresent their age as 13 or 40 or any other age—without so much as asking for proof. This is 19 analogous to selling a teenager alcohol who has admitted to being under 21 but then promptly 20 changed his story. 21 333. The upshot is that, in a matter of seconds, and without age verification, identity 22 verification, or parental consent, children of all ages can create a Facebook or Instagram account, 23 and immediately become subject to the products' various addictive and harmful features. 450 24 334. There can be no serious debate about whether Meta has more effective age 25 verification tools at its disposal. Meta has internal age identification models, such as the 26 "teen_non_teen" model or the "dim_ig_age_prediction_adult_classifier," that can estimate a user's 27 450 28 Similarly, the absence of effective age verification measures means that adult users can claim to be children—with obvious dangers to the actual children on Meta's products. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 106 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 113 of 278 1 age.451 Although this tool could be used to identify when a user is under 13 (or, for that matter, if a 2 user is a teenager and should therefore be safeguarded from particularly injurious aspects of Meta's 3 products) Meta does not use this safeguard. 452 4 335. Perversely, Meta does employ age verification on Instagram—but only when a user 5 self-reports they are younger than 13. In that case, Meta provides a user with what amounts to an 6 appeal right: "if you believe we made a mistake, please verify your age by submitting a valid photo 7 ID that clearly shows your face and date of birth." 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 451 Haugen_00003463, at Haugen_00003463- Haugen_00003465; see also Ibrahim Mousa Al- 26 Zaubi, Assef Jafar, & Kadan Aljoumaa, Predicting customer's gender and age depending on mobile phone data, 6 Journal of Big Data 18 (Feb 19, 2029), 27 https://journalofbigdata.springeropen.com/articles/10.1186/s40537-019-0180-9 (discussing generally how a similar age prediction algorithm works). 28 452 Haugen_00003463, at Haugen_00003463-Haugen_00003465. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 107 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 114 of 278 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 336. That is, instead of asking users to prove they are really over 13, Meta asks them if 20 they are really sure they are under 13. At best, this reflects a completely upside-down view of 21 Meta's duty of care, using age verification to screen in minor users but not to screen them out. At 22 worst, Meta's "are you sure you're really under 13" question invites pre-teens to falsify their 23 identification to gain access to Instagram. 24 337. Similarly, Meta imposes unnecessary barriers to the removal of accounts created by 25 children under 13. Since at least April 2018, Instagram and Facebook both accept reports of 26 accounts created by children under 13.453 However, before an Instagram or Facebook account is 27 453 Report an Underage User on Instagram, Instagram, 28 https://help.instagram.com/contact/723586364339719?fbclid=IwAR3E5rZo8zvp9Uw3giRoQRM PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 108 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 115 of 278 1 deleted, Meta requires verification that the child is under the age of 13. For example, Instagram's 2 reporting page states: 3 if you're reporting a child's account that was made with a false date of birth, and the child's age can be reasonably verified as under 13, 4 we'll delete the account. You will not get confirmation that the account has been deleted, but you should no longer be able to view 5 it on Instagram. Keep in mind that complete and detailed reports (example: providing the username of the account you're reporting) 6 help us take appropriate action. If the reported child's age can't reasonably be verified as under 13, then we may not be able to take 7 action on the account.454 8 Facebook's reporting page contains almost identical language. 455 By choosing to implement age 9 verification only before deleting accounts of users suspected to be children, but not when those 10 accounts are first created, Meta makes it more difficult to prove a user is under age 13 than it does 11 for a minor to pretend to be over 13. 12 338. It is unclear how long Meta takes to delete a reported account on average, if it does 13 so at all. Meta has ignored some parents' attempts to report and deactivate accounts of children 14 under 13 years old. 15 339. Zuckerberg has stated that he believes children under 13 should be allowed on 16 Facebook,456 so Meta's lax approach to age verification is no surprise. 17 340. Meta's approach to underage users of its product has consistently been one of 18 feigned ignorance. On October 10, 2021, Senator Marsha Blackburn reported that a young celebrity 19 told Instagram CEO Adam Mosseri that she had been active on Instagram since she was eight. 20 Mosseri replied that he "didn't want to know that." 457 21 y5qFmIGpy-NOLLtpctHOwkalXtfJ1ft9O09Q; Report an Underage Child, Facebook, 22 https://www.facebook.com/help/contact/209046679279097. 454 Report an Underage User on Instagram, Instagram, 23 https://help.instagram.com/contact/723586364339719?fbclid=IwAR3E5rZo8zvp9Uw3giRoQRM y5qFmIGpy-NOLLtpctHOwkalXtfJ1ft9O09Q. 24 455 Reporting an Underage Child, Facebook, https://www.facebook.com/help/contact/209046679279097. 25 456 Kashmir Hill, Mark Zuckerberg Is Wrong About Kids Under 13 Not Being Allowed on Facebook (May 20, 2011), https://www.forbes.com/sites/kashmirhill/2011/05/20/mark- 26 zuckerberg-is-wrong-about-kids-under-13-not-being-allowed-on-facebook/?sh=2ea85e825506. 457 27 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021), available at 28 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 109 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 116 of 278 1 341. But Meta does know that its age-verification protocols are inadequate to keep minors 2 off Facebook and Instagram. According to a May 2011 ABC News report, "about 7.5 million 3 [Facebook] users in the U.S. are under the age of 13, and about 5 million are under the age of 10." 458 4 Meta knows through retrospective cohort analyses that "up to 10 to 15% of even 10-year-olds in a 5 given cohort may be on Facebook or Instagram." 459 6 342. Meta knows that its chosen method of registration does not adequately protect minor 7 users from reporting inaccurate and implausible age information. As one product engineer 8 cautioned while analyzing the age of Facebook users, "Don't believe anything in the stated age 9 graph for under 30. They are all mixed up … We have way more people who say they are born in 10 the early 90's than exist in the population."460 11 343. Meta's internal studies confirm its knowledge that kids, tweens, and teens use its 12 products. In one study, Meta researched children as young as seven and found that in the fifth grade, 13 "social media becomes a part of their digital diet."461 Moreover, they identified that 24% of children 14 aged 7-9 and 38% of tweens aged 10-12 have at least one social media account, 462 and specifically 15 stated that Instagram's perceived user base included middle schoolers. 463 16 344. Another internal post reveals Meta's knowledge of the widespread use of Instagram 17 by preteens, as well as its targeting of children under the age of 13. In a study from around January 18 2021, titled "The Role of the Teen in Shaping a Household's Experience of Instagram," Meta 19 expressed a desire to utilize teenagers as the doorway into capturing an entire household of users, 20 including children under age 13.464 The post explains that teens can be used to teach their preteen 21 22 458 Ki Mae Heussner, Underage Facebook Members: 7.5 Million Users Under Age 13, ABC (May 9, 2011), https://abcnews.go.com/Technology/underage-facebook-members-75-million- 23 users-age-13/story?id=13565619. 459 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. 24 On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021), available at https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro 25 m%20a%20facebook%20whistleblower. 460 26 Haugen_00012303 at Haugen_00012314. 461 Haugen_00023849 at Haugen_00023910. 27 462 Haugen_00023849 at Haugen_00023866. 463 Haugen_00023849 at Haugen_00023879. 28 464 Haugen_00016728 at Haugen_00016728. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 110 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 117 of 278 1 siblings how to join while underage and to help them develop a habit of using and posting 2 indiscriminately.465 The article expresses concern that some teens may teach their preteen siblings 3 to post less and recommends that Meta combat this by changing perceptions among teens, so that 4 they will instruct their preteen siblings to use Instagram more spontaneously. 466 Key discussion 5 points from this document include: 6 Teens strongly influenced preteens' understanding of what and how frequently to share on IG, even discouraging them from sharing . . . 7 . We need to understand IG myths circulating among teens to inform comms and shift the perception of sharing on IG. . . . 8 Historically, teens have been a key focus for IG. Acquiring and 9 maintaining them continues to be a priority, reflected by investment in new features like Reels. Additionally, capturing the teen user 10 cohort on IG is critical as we think about Instagram's role within the broader family of apps. . . . [Teens] are typically the first in a 11 household to join. In many cases they're also critical to the onboarding process of parents and preteens alike. . . . 12 Older teens were IG catalysts for preteens. Most preteens became 13 curious about and wanted an IG account because of their older sibling. In some cases, preteens even relied on their older sibling to 14 create and set up their account, seeking their guidance on a username, profile, and accounts to [F]ollow. . . . If we're looking to acquire (and 15 retain) new users we need to recognize a teen's influence within the household to help do so, and the potential ripple effect. . . . 467 16 17 345. Meta has not used its concerning knowledge about preteen use and engagement to 18 comply with COPPA. Far to the contrary, it has leveraged its research to manipulate households 19 and target preteens through their siblings. 20 e. Facebook's and Instagram's parental controls are defective. 21 346. Facebook and Instagram lack adequate parental controls, which hinders parents' 22 ability to monitor and protect their children from harm. 23 347. Despite its obligations under COPPA, Meta does not require "verifiable parental 24 consent" for minors to use Facebook or Instagram. Meta has chosen to avoid its obligations by 25 26 27 465 Haugen_00016728 at Haugen_00016728-Haugen_00016732. 466 Haugen_00016728 at Haugen_00016736-Haugen_00016740. 28 467 Haugen_00016728 at Haugen_00016728- Haugen_00016734. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 111 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 118 of 278 1 purporting to ban children younger than 13, despite knowing that such children continue to access 2 and use its products due to its inadequate age verification methods. 3 348. Meta's lackadaisical approach to complying with COPPA is palpable in its internal 4 documents. To Meta, "the COPPA line is simple: we treat 13+ like other users and ask younger 5 people not to use our products."468 6 349. While COPPA requires parental consent only for users under the age of 13, a 7 reasonable company that knows or should have known its products are harmful to adolescents 8 would require parental consent for any minor to use them. But Meta's lack of parental consent 9 requirement for any underage users robs parents of an important way to protect their children from 10 the harms caused by Instagram and Facebook. 11 350. Those apps largely lack parental controls, despite their ready availability, 12 affordability, and ease of implementation. For example, Meta has chosen not to: (a) require 13 children's accounts on Facebook and Instagram to be linked to their parents', as it does with another 14 one of its products, Messenger Kids;469 (b) send reports of a child's activity to parents; (c) allow 15 parents to implement maximum daily usage limitations or to prohibit use during certain hours 16 (school, sleep hours, etc.); (d) notify parents about interactions with accounts associated with 17 adults; (e) notify parents when CSAM is found on a minor's account; or (f) require parental 18 approval before a minor can follow new accounts. 19 351. Controls like these would enable parents to track the frequency, time of day, and 20 duration of their child's use, identify and address problems arising from such use, and better 21 exercise their rights and responsibilities as parents. It is reasonable for parents to expect that social 22 media companies that actively promote their products to minors will undertake reasonable efforts 23 to notify parents when their child's use becomes excessive, occurs during sleep time, or exposes 24 the child to harmful content. Meta could feasibly design Instagram and Facebook to do so at 25 negligible cost. 26 468 Haugen_00017238 at Haugen_00017240. 27 469 Loren Chang, Introducing Messenger Kids, a New App for Families to Connect, Meta (Dec. 4, 28 2017), https://about.fb.com/news/2017/12/introducing-messenger-kids-a-new-app-for-families-to- connect/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 112 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 119 of 278 1 352. Meta creates a foreseeable risk to Plaintiffs through its defective products and then 2 attempts to shift the burden of protection from those products onto parents. As troublingly, Meta 3 intentionally designs Facebook and Instagram so that children can easily evade their parents' 4 supervision. Instagram and Facebook allow children to create a limitless number of anonymous 5 accounts without parental approval or knowledge, and also allows kids to block their parent's 6 profile.470 On Instagram, children can post stories to "Close Friends Only" (i.e., to a select group 7 of followers), excluding their parents. On Facebook, children can place their parents on a "restricted 8 list" of people who are unable to view their stories. 9 353. Meta has intentionally designed many aspects of Instagram and Facebook to 10 undermine parental supervision in an effort to maximize teen usage: "If Mom starts using an app 11 all the time, the app can lose a 'cool' factor, if we're not conscious of separation." 471 "We should 12 be thinking about how parents being on Instagram might effect graph management and teen 13 engagement over time. Discovery/usage of additional accounts could prove critical for authentic 14 sharing by teens."472 15 354. As one internal document described the issue: 16 [A]re teens able to maintain spaces that feel sacred to them (and their friends) or do we see decreased usage or new behavior patterns 17 emerge as household members join? . . . Preservation of protected spaces with require: [1] Learning how to create spaces within the app 18 where teens feel like they have privacy from both their own parents but also privacy from non-peers (e.g. Aunt Sally, neighbor down the 19 street, teachers, etc.). [2] Finding opportunities, such as ["]close friends["] where teens have their own, protected peer communities. 20 21 470 In 2018, Meta observed that "the participation rate of multiple account switching (basically the equivalent of Finstas) [was] going up," with 36% of teens engaging in multiple account 22 switching. Haugen_00017698 at Haugen_00017784. "Finsta," a widely used slang term, is a contraction of "fake" and "Insta" (short for Instagram). Caity Weaver and Danya Issawi, 'Finsta,' 23 Explained, N.Y. Times (Sept. 30, 2021), https://www.nytimes.com/2021/09/30/style/finsta- instagram-accounts-senate.html. "It is neither an official designation nor a type of account offered 24 by Facebook. Rather, it is a term many users ascribe to secondary accounts they create for themselves on Instagram, where their identities — and, often, the content of their posts — are 25 obscured to all but a small, carefully chosen group of followers." Caity Weaver and Danya Issawi, 'Finsta,' Explained, N.Y. Times (Sept. 30, 2021), 26 https://www.nytimes.com/2021/09/30/style/finsta-instagram-accounts-senate.html. 471 Haugen_00016728 at Haugen_00016735. 27 472 Haugen_00011969 at Haugen_00011974-75. "Graph management" apparently refers to efforts 28 by a user to unfollow accounts, i.e. "prun[e]." META3047MDL-003-00146492 at META3047MDL-003-00146495; META3047MDL-003-00178437. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 113 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 120 of 278 1 [3] Understanding the value of certain features being more complex (i.e. indirectly made for teens because more challenging for parents 2 or preteens). Both snapchat and TikTok are somewhat confusing to parents, in turn affording teens a protected place to play/engage. 473 3 355. Meta's internal documents recognize that parents are largely ill-equipped to protect 4 children from its products. For example, one employee asserted in the discussion of a focus group 5 survey regarding the mental health impact of Meta's products on teenagers: 6 The other big reason that parents are not a source of support has to 7 do with parents' ability (or really, their inability) to understand what adolescence in the age of social media looks and feels like. The 8 parents of today's teens came of age before social media, so they don't know and *can't* what it's like to live in what feels like a 9 constant spotlight. When today's parents were teens, social comparison was much more limited both in terms of scope and scale. 10 Teens today compare themselves to many more people, much more often, and about more parts of life than their parents did during their 11 adolescence. In addition, today's parents were able to turn it off when they went home, while teens feel compelled to be on social media all 12 the time.474 13 356. When employees have raised the possibility of additional safeguards—"could we 14 offer a parental control feature so that parents and kids could learn and cope together?"—Meta has 15 consistently ignored them.475 16 357. Finally, Meta has failed to develop effective reporting tools to deal with abuse 17 directed at children through Instagram and Facebook. Meta does not have a phone number that a 18 parent or child can call to report such abuse in real time. Its online reporting mechanisms lack an 19 immediate response mechanism, regardless of the seriousness of the harm at issue. And certain 20 Plaintiffs have found that Meta declined to respond to reports filed through its online reporting tool, 21 citing technical issues. 22 23 24 25 26 27 473 Haugen_00016728 at Haugen_00016735. 474 Haugen_00017069 at Haugen_00017173. 28 475 Haugen_00017069 at Haugen_00017173. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 114 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 121 of 278 1 f. Facebook's and Instagram's defective features include impediments to discontinuing use. 2 3 358. Meta has intentionally and defectively designed its products so that adolescent users, 4 including Plaintiffs, face significant navigational obstacles and hurdles when trying to delete or 5 deactivate their accounts, in contrast to the ease with which users can create those accounts. 6 359. Currently, to delete or deactivate an Instagram or Facebook account, a user must 7 locate and tap on approximately seven different buttons (through seven different pages and popups) 8 from the main feed. Some Plaintiffs have given up in their attempt to quit because it was too difficult 9 to navigate through the interface to completion. 10 360. Even if a user successfully navigates these seven pages, Meta still won't 11 immediately delete their account. Instead, Meta preserves the account for 30 more days. If at any 12 time during those 30 days a user's addictive craving becomes overwhelming and they access the 13 account again, the deletion process starts over. The user must go through all the above steps again, 14 including the 30-day waiting period, if they again wish to delete their account. 15 16 17 18 19 20 21 22 23 24 25 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 115 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 122 of 278 1 361. Moreover, the deletion process includes what Meta readily acknowledges are 2 "aggressive" attempts to dissuade users from deleting their accounts. 476 Before a user can delete 3 their Facebook account, Meta "lists some of your friends to remind you that they will no longer be 4 able to contact you through the site and more importantly, [requires] the user to choose a reason for 5 why they're leaving."477 Meta also requires users attempting to leave Instagram to select a reason 6 why they are leaving. 7 362. As an additional barrier to deletion, Meta urges users of both products to deactivate, 8 rather than delete, their accounts. For example, Instagram users who choose to delete their accounts 9 are immediately shown a screen with their profile picture and asked: "Deactivate your account 10 instead of deleting?" The option to deactivate is conspicuously highlighted. Similarly, Facebook 11 displays a screen that automatically selects the option of deactivating rather than deleting a user 12 account. 13 363. Meta's aggressive efforts to prevent users from discontinuing their use of Facebook 14 and Instagram is particularly problematic because unsuccessful efforts to discontinue use are a 15 hallmark of addiction, incorporated as the sixth criteria in the Bergen Social Media Addiction Scale, 16 discussed above. 17 4. Meta has concealed from Plaintiffs, the public, and Congress the harmful effects that Instagram's and Facebook's design have on 18 children. 19 364. Meta has engaged in a years-long pattern of concealing critical information about 20 the safety of Instagram and Facebook from the public, including Plaintiffs and their parents. In one 21 internal document from February 2018, employees at Meta communicated internally about how 22 best to "refin[e] counter-messaging around the addiction narrative that's been propagating." This 23 effort to conduct "message testing around addiction PR responses" included the ideas that "[t]he 24 whole dopamine thing is completely made up and based on no research," "[t]here's no agreement 25 26 27 476 Haugen_00016893 at Haugen_0001689398. 28 477 Haugen_00016893 at Haugen_0001689398. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 116 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 123 of 278 1 on what is meant by addiction," and (contradictorily) "[w]e're taking it seriously, doing research, 2 [and] launching new tools to help people." 478 3 365. Meta knew that none of this was true. For instance, in the summer of 2019, 4 Zuckerberg met with a psychologist and leading expert on the mental health effects of social media 5 on young people. This leading expert countered Zuckerberg's contention that harms from social 6 media are trivial and explained how, to the contrary, Instagram and other products have been a 7 major contributor to the spike in young girls' mental health problems since 2012. The psychologist 8 addressed his research "on the dramatic rise in rates of teenage anxiety, depression, and self-harm" 9 and explained how the research on social media's role "points heavily to a connection, not just from 10 correlational studies but from true experiments, which strongly indicate causation, not just 11 correlation."479 12 366. Instead of "taking [this] seriously" and "launching new tools" to protect kids,480 13 Meta did the opposite. By late 2019, Meta's "mental health team stopped doing things," "it was 14 defunded" and "completely stopped."481 And, as noted, Meta allowed safety tools it knew were 15 broken to be held out as fixes.482 All the while, Meta ignored cries from their well-being researchers 16 to aggressively confront its youth safety problem: "there's so much more we could have done 17 here … [but] there was the explicit decision last half not to fund this anymore." 483 18 367. Despite knowing better, Meta's high-ranking executives then began pushing 19 intentionally misleading talking points to the public. Instead of informing the public about Meta's 20 internal research demonstrating Instagram's and Facebook's negative impacts on the health and 21 well-being of the nation's youth, Meta repeatedly omitted key facts and misrepresented its products 22 in service of an overall message touting the safety of its products for children. 23 24 478 META3047MDL-003-00082165 at META3047MDL-003-00082165- META3047MDL-003- 00082165. 25 479 META3047MDL-003-00089174 at META3047MDL-003-00089176. 480 26 META3047MDL-003-00082165 at META3047MDL-003-00082165- META3047MDL-003- 00082165. 27 481 META3047MDL-003-00011697 at META3047MDL-003-00011698. 482 See supra ¶ 302. 28 483 META3047MDL-003-00103260 at META3047MDL-003-00103260. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 117 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 124 of 278 1 368. Because of Meta's concealment, Plaintiffs, Consortium Plaintiffs, the public, and 2 Congress were left in the dark and reasonably relied on Meta's reassurances. Had Meta disclosed 3 the truth regarding its products, Plaintiffs and Consortium Plaintiffs would have been able to avoid 4 or mitigate, and many would have avoided and mitigated, the harms they ultimately suffered by 5 using Meta's products. Instead, Meta pursued a knowing pattern of concealment to Plaintiffs' 6 detriment. 7 369. In the year leading up to Meta's acquisition of Instagram, Meta publicly 8 acknowledged its duty to children and worked to create false expectations about its products' safety. 9 For example, 10 Zuckerberg (3/25/2011): "So, we're really focused on, on safety, especially 11 children's safety. So we're having folks under the age of 18, um we, we just 12 take a lot of extra precautions for it, to make sure that it's just a safe 13 environment for them um, to use this service that you know, the default for, 14 for people sharing things isn't that they're sharing with everyone but that 15 they're sharing with a smaller community … But I think, I think that's a lot 16 of it. We really try to build a safe environment. Um, and um, that's gonna be 17 the key long term."484 18 Zuckerberg (3/25/2011): "Right, and they, they feel like Facebook is this 19 really secure place and that it's a hundred percent safe, and um, we're always 20 thinking about little and big things like that that we can do to keep it safe for, 21 for the people who use our service."485 22 Zuckerberg (5/25/2011): "I mean, we do not allow people under the age of 23 13 to sign up and I think if we ever were, we would need to try to figure out 24 a lot of ways to make sure that they were safe, right, because that's just 25 26 484 Mark Zuckerberg at BYU with Senator Orrin Hatch, YouTube, March 25, 2011, 27 http://www.youtube.com/watch?v=zRsbWOmmvNo. 485 28 Mark Zuckerberg at BYU with Senator Orrin Hatch, YouTube, March 25, 2011, http://www.youtube.com/watch?v=zRsbWOmmvNo. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 118 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 125 of 278 1 extremely important and that's just not the top of the list in terms of things 2 for us to figure out right now." 486 3 370. Following Meta's acquisition of Instagram, high-ranking executives continued to 4 make public pronouncements about the safety of Meta's products, including but not limited to the 5 following statements: 6 Zuckerberg (12/1/2015): "We will do our part to make this [better world] 7 happen, not only because we love you, but also because we have a moral 8 responsibility to all children in the next generation."487 9 Zuckerberg (4/11/2018): "Congressman, we have a number of measures in 10 place to protect minors specifically. We make it so that adults can't contact 11 minors who they - they aren't already friends with. We make it so that certain 12 content that may be inappropriate for minors, we don't show." 488 13 Zuckerberg (4/10/2018): when asked by members of the U.S. Senate 14 Committee on Commerce, Science, and Transportation whether his 15 companies "[h]ire consulting firms to help them figure out how to get more 16 dopamine feedback loops so that people don't want to leave the platform": 17 "No . . . that's not how we talk about this or how we set up our product teams. 18 We want our products to be valuable to people, and if they're valuable, then 19 people choose to use them."489 20 Zuckerberg (7/12/2018): "There are really two core principles at play here. 21 There's giving people a voice, so that people can express their opinions. 22 486 Maurice Levy, Conversation with Mark Zuckerberg at E-G8 Forum, YouTube, May 25, 2011, 23 http://www.youtube.com/watch?v=Gy0bq9FAJRs. 487 Mark Zuckerberg, Facebook (Dec. 1, 2015), 24 https://www.facebook.com/zuck/posts/10153375081581634. 488 Transcript of Zuckerberg's appearance before House committee, Wash. Post (April 11, 2018) 25 https://www.washingtonpost.com/news/the-switch/wp/2018/04/11/transcript-of-zuckerbergs- appearance-before-house-committee/?utm_term=.e7b476fb8ac7&noredirect=on. 26 489 Facebook, Social Media Privacy, and the Use and Abuse of Data: Hearing Before the S. 27 Comm. on Commerce, Sci., and Transp. and H. Comm's on the Judiciary and Commerce, Sci., and Transp., 115th Cong. (Apr. 10, 2018), available at 28 https://www.commerce.senate.gov/2018/4/facebook-social-media-privacy-and-the-use-and- abuse-of-data. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 119 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 126 of 278 1 Then, there's keeping the community safe, which I think is really 2 important."490 3 Zuckerberg (7/25/2018): "[W]e will continue to invest heavily in security 4 and privacy because we have a responsibility to keep people safe. But as I've 5 said on past calls, we're investing so much in security that it will 6 significantly impact our profitability."491 7 Zuckerberg (8/21/2018): "One of the most important responsibilities we 8 have as a company is to keep people safe and stop anyone from abusing our 9 service."492 10 Zuckerberg (9/7/2018): "What I've learned so far is that when you build 11 services that are used by billions of people across countries and cultures, you 12 will see all of the good humanity is capable of, and people will try to abuse 13 those services in every way possible. It is our responsibility to amplify the 14 good and mitigate the bad."493 15 Zuckerberg (11/15/2018): "[W]e have a responsibility to keep people safe 16 on our services -- whether from terrorism, bullying, or other threats." 494 17 18 19 20 490 Kara Swisher, Zuckerberg: The Record Interview, Vox (July 12, 2018) 21 https://www.vox.com/2018/7/18/17575156/mark-zuckerberg-interview-facebook-recode-kara- swisher. 491 22 Facebook, Inc., Second Quarter 2018 Results Conference Call (July 25, 2018) https://s21.q4cdn.com/399680738/files/doc_financials/2018/Q2/Q218-earnings-call- 23 transcript.pdf. 492 Mark Zuckerberg, Facebook (Aug. 21, 2018), 24 https://www.facebook.com/zuck/posts/10105188590724391?__tn__=K-R. 493 Mark Zuckerberg, Facebook (Sept. 7, 2018), 25 https://www.facebook.com/zuck/posts/10105224999156601?__xts__%5B0%5D=68.ARB273c8T JkMqNAclfl- 26 i0UB6fVWHZ_hO4k0KASCy8XfVdyC9XEVqoPLsPUPDh94zSHboQiB1t3mSlP9yEUyjvaEF5 0UxoUqVca4ZcM4nnkQ3MWz3dBGRQYm7lJMj_Cbl25p7a9-HX- 27 aXjkjNdS21XzaAThg9PfkrzJ_dTLszwUZ3H6b3Q4biIc&__tn_. 494 28 Mark Zuckerberg, Facebook (Nov. 15, 2018), https://www.facebook.com/notes/mark- zuckerberg/a-blueprint-for-content-governance-and-enforcement/10156443129621634/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 120 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 127 of 278 1 Zuckerberg (1/1/2019): "We ended 2018 with more than 30,000 people 2 working on safety and security -- up from 10,000 people a couple of years 3 ago."495 4 Zuckerberg (1/30/2019): "[O]n all the content and safety and security issues, 5 there's more to do here but I'm proud of the work that we have done to get 6 in front of a lot more of these issues."496 7 Zuckerberg (3/30/2019): "[W]e have a responsibility to keep people safe on 8 our services."497 9 Zuckerberg (4/24/2019): "You should expect we'll do everything we can to 10 keep you safe on our services, within the bounds of an encrypted service." 498 11 Sheryl Sandberg (1/29/2020): "[We] have to keep people safe and give them 12 control over their experience on our apps. And we are." 499 13 Sheryl Sandberg (10/29/2020): "While we continue to invest in helping 14 businesses, we are equally focused on keeping our platform safe." 500 15 Meta (12/23/2020), when asked by the Senate Committee on the Judiciary 16 whether it could "determine whether increased use of their platform among 17 teenage girls has any correlation with increased signs of depression [or 18 anxiety]": "No." And, when asked what research Meta had conducted 19 20 495 Meta Investor Relations, Earnings Call Transcript, Meta (Jan. 1, 2019), 21 https://investor.fb.com/investor-events/event-details/2019/Facebook-Q4-2018- Earnings/default.aspx. 22 496 Meta Investor Relations, Earnings Call Transcript, Meta (Jan. 30, 2019), https://investor.fb.com/financials/default.aspx. 23 497 Mark Zuckerberg, Mark Zuckerberg: The Internet needs new rules. Let's start in these four 24 areas, Wash. Post (March 30, 2019), https://www.washingtonpost.com/opinions/mark- zuckerberg-the-internet-needs-new-rules-lets-start-in-these-four-areas/2019/03/29/9e6f0504- 25 521a-11e9-a3f7-78b7525a8d5f_story.html?noredirect=on. 498 Mark Zuckerberg, Facebook (April 24, 2019), 26 https://www.facebook.com/zuck/posts/10107243286682221. 499 Meta Investor Relations, Earnings Call Transcript, Meta (Jan. 29, 2020), 27 https://investor.fb.com/investor-events/default.aspx. 500 Meta Investor Relations, Earnings Call Transcript, Meta (Oct. 29, 2020), 28 https://investor.fb.com/investor-events/default.aspx. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 121 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 128 of 278 1 internally on the mental health impacts of social media use: "[t]he effects of 2 social media are still being studied."501 3 Zuckerberg (3/25/21), when asked by members of the U.S. House of 4 Representatives Committee on Energy and Commerce, "Do you believe that 5 your platform harms children?": "I don't believe so. This is something that 6 we study and we care a lot about; designing products that improve peoples' 7 well-being is very important to us. And what our products do is help people 8 stay connected to people they care about, which I think is one of the most 9 fundamental and important human things that we do, whether that's for teens 10 or for people who are older than that."502 11 David Wehner, Chief Financial Officer of Meta (4/28/2021): "I mean, the 12 only thing I'd add . . . is that, I think more than anyone else in the industry, 13 we invest on the safety and security side to sort of keep bad content off the 14 site before it gets ranked and put into what people see. So we've got 35,000 15 -- over 35,000 people on the safety and security side. We've got the most 16 robust set of content policies out there. We do a quarterly call, public call 17 around our content review process and procedures. So I think that on the 18 front, before it even gets into the algorithm, I think we really do more than 19 anyone else in the industry on the safety and security front to prevent things 20 like misinformation and a bad content going into the system in the first 21 place."503 22 501 Facebook, Inc. Responses to Questions for the Record from the Comm. on the Judiciary 23 November 17, 2020 Hearing: Breaking the News: Censorship, Suppression, and the 2020 Election, at 124-125 (December 23, 2020), available at 24 https://www.judiciary.senate.gov/imo/media/doc/Zuckerberg%20Responses%20to%20QFRs.pdf. 502 Disinformation Nation: Social Media's Role in Promoting Extremism and Misinformation 25 Hearing Before H. Energy and Commerce Subcomm. on Communications and Technology (March 25, 2021), available at 26 https://www.congress.gov/117/meeting/house/111407/documents/HHRG-117-IF16-Transcript- 20210325.pdf. 27 503 Meta Investor Relations, Earnings Call Transcript, Meta (April 28, 2021), 28 https://investor.fb.com/investor-events/event-details/2021/Facebook-Q1-2021-Earnings- /default.aspx. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 122 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 129 of 278 1 Adam Mosseri (5/2021): in statement to reporters, dismissing concerns 2 around Instagram's negative impact on teens as "quite small." 504 3 371. On each of the above occasions, and on many others, Meta touted the safety of its 4 products; it could have but failed to disclose information it knew concerning the significant risks 5 associated with its products, even though it knew that the public lacked access to this information. 6 For instance, in a December 2019 memo, Meta's Chief Technology Officer, remarked that the 7 media has "limited information to work with" about the company and that this limitation is by 8 Meta's "own design."505 9 372. Meta's pattern of intentional concealment came to a head in August 2021, just weeks 10 before Frances Haugen dropped her bombshell revelations on the public. On August 4, 2021, 11 Senators Marsha Blackburn and Richard Blumenthal wrote to Mark Zuckerberg. The Senators' 12 letter observed that "[a]n expanding volume of scientific research shows that social media platforms 13 can have a profoundly harmful impact on young audiences" and noted "grave concerns about 14 [Meta's] apparent effort to ensnare children into social media platforms at earlier and earlier 15 ages."506 The letter concluded by asking Zuckerberg six "pretty straightforward questions about 16 how the company works and safeguards children and teens on Instagram." 507 17 18 19 20 504 Taylor Hatmaker, Facebook Knows Instagram Harms Teens. Now its Plan to Open the App to 21 Kids Looks Worse than Ever, TechCrunch (Sept. 16, 2021), available at https://techcrunch.com/2021/09/16/facebook-instagram-for-kids-mosseri-wsj-teen-girls/. 22 505 Haugen_00007350 at Haugen_00007350 (Dec. 30, 2019 memo by Andrew Bosworth regarding "Thoughts for 2020"). 23 506 Letter from Richard Blumenthal, U.S. Senator, to Mark Zuckerberg, Chief Executive Officer of Facebook (Aug. 4, 2021), available at 24 https://www.blumenthal.senate.gov/imo/media/doc/8.4.21%20-%20Facebook%20- %20Mental%20Health%20and%20Kids%20Letter.pdf. 25 507 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. 26 On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021), available at https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro 27 m%20a%20facebook%20whistleblower. See also, Letter from Richard Blumenthal, U.S. Senator, to Mark Zuckerberg, Chief Executive Officer of Facebook (Aug. 4, 2021), available at 28 https://www.blumenthal.senate.gov/imo/media/doc/8.4.21%20-%20Facebook%20- %20Mental%20Health%20and%20Kids%20Letter.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 123 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 130 of 278 1 373. In its August 17, 2021 written response to Senators Blackburn and Blumenthal, 2 Meta omitted any reference to the internal research it had conducted demonstrating the negative 3 impact Instagram can have on kids' mental health. 508 4 374. The Senators' letter asked whether Meta had ever developed products or features 5 "that it had reason to believe could have a negative effect on children's and teens' mental health or 6 well-being."509 Meta responded by claiming it had "built many special protections for teens." 510 7 But it failed to mention, for example, that it employed "growth hackers" who internally advised, 8 "we can be very aggressive with our notifications to create a habit." 511 9 375. The Senators' letter also asked if Meta's research had "ever found that its platforms 10 and products can have a negative effect on children's and teens' mental health or well-being." 512 11 Meta responded that the matter was "still being studied," 513 that it was challenging to conduct such 12 research,514 and that the company was "not aware of a consensus among studies or experts about 13 how much screen time is 'too much.'"515 While Meta reiterated its vague and already public 14 15 16 508 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. 17 Senator (Aug. 17, 2021), https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf. 18 509 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. Senator 4 (Aug. 17, 2021), 19 https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf. 510 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. 20 Senator 4 (Aug. 17, 2021), https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf. 21 511 Haugen_00016893 at Haugen_00016914 (Aug. 3, 2017 memo entitled "Have we made people 22 addicted to Facebook?"). 512 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. 23 Senator 2 (Aug. 17, 2021), https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf. 24 513 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. Senator 2 (Aug. 17, 2021), 25 https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf. 514 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. 26 Senator 3 (Aug. 17, 2021), https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf. 27 515 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. 28 Senator 3 (Aug. 17, 2021), https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 124 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 131 of 278 1 position that "passive" use of social media can correlate with "negative outcomes," 516 it failed to 2 disclose any more specific findings.517 3 376. Meta should have, but intentionally did not, respond to the Senators' question by 4 disclosing its detailed research regarding addiction to its products, which the company terms 5 problematic usage;518 its assessment that "[t]he best external research indicates that Facebook's 6 impact on people's well-being is negative";519 its identification of "Problematic Use," loneliness, 7 and social comparison as the three drivers of this negative impact; 520 its finding that up to 25% of 8 people on Facebook experience so-called problematic use;521 its data showing that "high time spent 9 users do tend to be disproportionately younger users"; 522 its conclusion that so-called problematic 10 use causes profound harms, including loss of productivity, sleep disruption, relationship impacts, 11 and safety risks;523 its identification of multiple Meta product features that act as triggers for so- 12 13 516 14 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. Senator 3 (Aug. 17, 2021), 15 https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf.; see also Meta Investor Relations, Earnings Call Transcript, Meta (April 25, 2018), 16 https://s21.q4cdn.com/399680738/files/doc_financials/2018/Q1/Q1-18-Earnings-call- transcript.pdf. 17 517 Letter from Facebook, Inc. to Richard Blumenthal, U.S. Senator, and Marsha Blackburn, U.S. Senator 6 (Aug. 17, 2021), 18 https://www.blumenthal.senate.gov/imo/media/doc/817.21facebookresponseletter.pdf. 518 Haugen_00016373 at Haugen_00016379 (Mar. 9, 2020 internal presentation and discussion 19 about problematic use with a slide stating that problematic use "is sometimes referred to as 'social media addiction' externally"); Haugen_00016373 at Haugen_00016373 (Mar. 9, 2020 internal 20 presentation and discussion regarding problematic use in which a Meta employee shared a post stating: "In Q4 2019, our Well-being Product Team conducted global qualitative research to 21 better understand 'problematic' use (sometimes called 'social media addiction' externally"); Haugen_00005458 at Haugen_00005473 (Nov. 5, 2019 report by Meta employee regarding 22 "Hard Life Moments – Mental health deep dive"); Haugen_00007055 at Haugen_00007055 (May 6, 2019 memo by Meta employee regarding "Problematic use / time-spent papers at CHI"). 23 519 Haugen_00016373 at Haugen_00016381 (Mar. 9, 2020 internal presentation and discussion 24 about problematic use). 520 Haugen_00016373 at Haugen_00016381. 25 521 Haugen_00016373 at Haugen_00016383. 522 Haugen_ 00017177 at Haugen_00017181 (Oct. 30, 2018 report by Meta employee regarding 26 "How Behavior on Instagram Varies with Overall Time Spent"); Haugen_00005458 at Haugen_00005750-Haugen_00005751 (Sept. 18, 2019 presentation containing slides about brain 27 maturation). 523 28 Haugen_00016373 at Haugen_00016414 (Mar. 9, 2020 presentation stating "All problematic users were experiencing multiple life impacts"). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 125 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 132 of 278 1 called problematic use;524 its knowledge that teens who feel addicted to a Meta app "know that 2 what they're seeing is bad for their mental health but feel unable to stop themselves"; 525 its studies 3 regarding body image and social comparison;526,527 its knowledge that Instagram makes body image 4 issues worse "for one in three teen girls";528 its analysis showing that topics eliciting appearance 5 comparison comprise one third of what teen girls see on Instagram; 529 its research concluding that 6 negative social comparison on Instagram gets worse for users over time; 530 its awareness that teens 7 report Instagram as a source of increased anxiety and depression; 531 its finding that Instagram has 8 a "consistent bias in favor of harmful content"; 532 its knowledge that Meta's recommendation 9 algorithms "create an echo chamber" of suicide and self-harm content; 533 its researchers' 10 conclusion that teens "[h]ave an addict's narrative about their use" of Instagram; 534 and its survey 11 finding that "[o]ver one third of teens felt they have only a little control of nor control at all over 12 how Instagram makes them feel" 535—in addition to the other findings described in this Complaint. 13 524 Haugen_00016373 at Haugen_00016410 ("We heard about 10+ triggers contributing to PU 14 habits"). 525 Haugen_00017069 at Haugen_00017171 (Oct. 10, 2019 report by Meta employee and 15 discussion about teens' mental health). 526 Haugen_00005458 at Haugen_00005484 (Sept. 18, 2019 presentation regarding "Mental 16 Health Findings"). 527 17 Haugen_00000797 at Haugen_00000797 (Nov. 16, 2018 report regarding "IG Social Comparison Research Findings"). 528 18 Haugen_00005458 at Haugen_00005500 (Sept. 18, 2019 presentation containing a slide stating "But, We Make Body Image Issues Worse for 1 in 3 Teen Girls"). 19 529 Haugen_00002527 at Haugen_00002527 (Mar. 9, 2021 report regarding "How the topics people see are linked to appearance comparison on IG"). 20 530 Haugen_00000797 at Haugen_00000875 (Nov. 16, 2018 report containing a page displaying data about negative social comparison over time). 21 531 Haugen_00017069 at Haugen_00017121 (Oct. 10, 2019 presentation containing a slide 22 regarding "Teens blame Instagram for increases in the rates of anxiety and depression among teens"). 532 23 Haugen_00003739 at Haugen_00003739 (Undated report regarding "Is Instagram Reels Favoring badness?"). 24 533 Haugen_00005378 at Haugen_00005379 (Dec. 2, 2020 report regarding "Tackle Community- Based Harm in Dangerous Content"). 25 534 Subcomm.: Protecting Kids Online: Facebook, Instagram, and Mental Health Harms Hearing before Subcomm. On Consumer Protection Product Safety, and Data Security (Sept. 30, 2021), 26 available at https://www.commerce.senate.gov/2021/9/protecting-kids-online-facebook- instagram-and-mental-health-harms ). 27 535 Subcomm.: Protecting Kids Online: Facebook, Instagram, and Mental Health Harms Hearing 28 before Subcomm. On Consumer Protection Product Safety, and Data Security (Sept. 30, 2021), PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 126 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 133 of 278 1 377. Meta's years-long concealment of its research was revealed just weeks later, when 2 Frances Haugen released these studies, along with a trove of other internal Meta documents, to the 3 Wall Street Journal. Even these disclosures did not reveal the full scope and extent of Meta's 4 misrepresentations, discussed elsewhere in this complaint. 5 378. On September 21, 2021, Senator Blumenthal confronted a Meta representative about 6 the conspicuous omissions in Meta's response to his letter: 7 Last month, on August 4, Senator Blackburn and I wrote to Mark Zuckerberg and asked him specifically about this issue. We asked, 8 and I'm quoting, "Has Facebook's research ever found that its platforms and products can have a negative effect on children's and 9 teens' mental health or well-being such as increased suicidal thoughts, heightened anxiety, unhealthy usage patterns, negative 10 self-image, or other indications of lower well-being?" 11 It wasn't a trick question. It preceded the reports in the Journal. We had no idea about the whistleblower documents that were ultimately 12 revealed. 13 Facebook dodged the question. "We are not aware of a consensus among studies or experts about how much screen time is too much." 14 We are not aware. Well, we all know now that representation was 15 simply untrue.536 16 379. Senator Blumenthal went on to ask the witness, Facebook's Vice President of 17 Privacy & Public Policy, "why did Facebook misrepresent its research on mental health and teens 18 when it responded to me and Senator Blackburn?" After disputing the characterization, Satterfield 19 responded, "The safety and well-being of the teens on our platform is a top priority for the company. 20 We're going to continue to make it a priority. This was important research." Senator Blumenthal 21 then went on: "Why did you conceal it?" Satterfield responded, "we didn't make it public because 22 23 24 available at https://www.commerce.senate.gov/2021/9/protecting-kids-online-facebook- 25 instagram-and-mental-health-harms. 536 26 Richard Blumenthal, Blumenthal Demands Facebook Appear at Next Week's Consumer Protection Subcomm. Hearing to Explain Coverup of its Platforms' Negative Impact on Teens 27 and Children (Sept. 21, 2021), available at https://www.blumenthal.senate.gov/newsroom/press/release/blumenthal-demands-facebook- 28 appear-at-next-weeks-consumer-protection-subcommittee-hearing-to-explain-coverup-of-its- platforms-negative-impact-on-teens-and-children. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 127 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 134 of 278 1 we don't, with a lot of the research we do because we think that is an important way of encouraging 2 free and frank discussion within the company about hard issues." 537 3 380. Meta unilaterally decided to prioritize "free and frank" internal discussion over 4 honest and transparent responses to direct questions from sitting United States Senators. When it 5 "dodged, ducked, sidetracked, [and] in effect misled" Senators Blumenthal and Blackburn, Meta 6 deceived the public via its elected representatives. 538 7 381. Moreover, Satterfield's "free and frank discussion" excuse has been contradicted 8 both internally and publicly by Meta employees. On January 8, 2020, a Meta software engineer 9 participated in an internal "ask me anything" session, on the last day of his four-year tenure at the 10 company. When asked how the Meta Defendants should respond to outside pressures and critiques, 11 that software engineer stated: "Right now, many employees feel that if they whistleblow, dissent, 12 give feedback to unethical decisions, etc, then they are at risk for being fired. We can fix that by 13 giving people the safety to speak up when they see something wrong going on." 539 14 382. Frances Haugen echoed this sentiment in her testimony before the Senate, citing 15 evidence that Meta "is so scared of even basic transparency that it goes out of its way to block 16 researchers who are asking awkward questions."540 Ms. Haugen further testified that Meta's culture 17 537 18 Richard Blumenthal, Blumenthal Demands Facebook Appear at Next Week's Consumer Protection Subcomm. Hearing to Explain Coverup of its Platforms' Negative Impact on Teens 19 and Children (Sept. 21, 2021), available at https://www.blumenthal.senate.gov/newsroom/press/release/blumenthal-demands-facebook- 20 appear-at-next-weeks-consumer-protection-subcommittee-hearing-to-explain-coverup-of-its- platforms-negative-impact-on-teens-and-children. 21 538 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021) available at 22 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower; see also Protecting Kids Online: Testimony from a 23 Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021) available at 24 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower (statement by Senator Brian Schatz to Frances Haugen 25 that he had "a long list of misstatements, misdirections and outright lies from the company"). 539 Haugen_00007481 at Haugen_00007492 (Jan. 8, 2020 report regarding "Political Ads 26 Announcement Preview [Confidential]"). 540 27 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021) available at 28 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 128 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 135 of 278 1 emphasizes insularity and promotes the idea that "if information is shared with the public, it will 2 just be misunderstood."541 3 383. The above representations of former employees are consistent with reports from 4 Facebook content moderators that there is a "culture of fear and excessive secrecy" within Meta 5 that "prevent[s] [them] from speaking out."542 6 384. Notably, Meta's pattern of concealment did not end after Frances Haugen came 7 forward. On September 30, 2021, Antigone Davis, Facebook's Head of Safety, testified before the 8 Senate. Ms. Davis represented that, when Instagram "do[es] ads to young people, there are only 9 three things that an advertiser can target around: age, gender, location. We also prohibit certain ads 10 to young people, including weight-loss ads." 543 She further testified, "We don't allow the 11 sexualization of minors on our platform." 544 12 385. Ms. Davis's statements were subsequently proven false by Senator Mike Lee. 13 During an October 2021 hearing, Senator Lee explained that a group called the Technology 14 Transparency Project ("TTP") alerted the U.S. Senate that it had gained Facebook's approval to 15 target a series of harmful ads to up to 9.1 million users between the ages of 13 and 17. 545 While 16 TTP did not actually run the ads, approval from Meta to do so demonstrates that the company 17 541 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. 18 On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021) available at https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro 19 m%20a%20facebook%20whistleblower. 542 20 Zoe Schiffer, Facebook Content Moderators Call for Company to Put an End to Overly Restrictive NDAs, The Verge (Jul. 22, 2021), available at 21 https://www.theverge.com/2021/7/22/22587757/facebook-content-moderators-ireland-end- restrictive-ndas. 22 543 Subcomm.: Protecting Kids Online: Facebook, Instagram, and Mental Health Harms Hearing before Subcomm. On Consumer Protection Product Safety, and Data Security (Sept. 30, 2021), 23 available at https://www.commerce.senate.gov/2021/9/protecting-kids-online-facebook- instagram-and-mental-health-harms. 24 544 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021) available at 25 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower. 26 545 See Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before 27 Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021) available at 28 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 129 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 136 of 278 1 allows harmful targeted advertising toward minors. Senator Lee showed three examples of these 2 Meta-approved ads, shown below:546, 547 3 4 5 6 7 8 9 10 11 12 13 386. The first ad encourages children to "[t]hrow a skittles party like no other" and 14 displays the suggestion against a background of colorful prescription pills. The second ad promotes 15 an "Ana Tip" instructing the viewer to "visit pro-ana sites to feed your motivation and reach your 16 goal" when feeling hungry. The third ad informs the viewer that they "look lonely" and encourages 17 them to "[f]ind your partner now to make a love connection." 18 387. Senator Lee stated that based on the Meta Defendants' approval of these pro-drug, 19 pro-anorexia, pro-sexualization ads targeted to children aged 13 to 17, "[o]ne could argue that it 20 21 22 23 546 These screen captures were taken from a video of the October 5, 2021 Senate Hearing with 24 witness Frances Haugen. See Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 25 2021) available at https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro 26 m%20a%20facebook%20whistleblower. 547 27 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021) available at 28 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 130 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 137 of 278 1 proves that Facebook is allowing and perhaps facilitating the targeting of harmful adult-themed ads 2 to our nation's children."548 3 388. In addition to the litany of misrepresentations and omissions identified above, Meta 4 has repeatedly failed to tell the truth about the age of users on Instagram. In statements to Congress 5 and elsewhere, Zuckerberg has represented that Meta does not allow users under the age of 13 to 6 use the product. For example, in testimony before the U.S. House of Representatives Committee 7 on Energy and Commerce, Zuckerberg stated: "There is clearly a large number of people under the 8 age of 13 who would want to use a service like Instagram. We currently do not allow them to do 9 that."549 10 389. However, as discussed further above, Meta has long known that its product is widely 11 used by children under the age of 13. In fact, Meta knows through retrospective cohort analyses 12 that "up to 10 to 15% of even 10 year-olds in a given cohort may be on Facebook or Instagram." 550 13 14 15 548 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021) available at 16 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower. 17 549 Disinformation Nation: Social Media's Role in Promoting Extremism and Misinformation Hearing Before H. Energy and Commerce Subcomm. on Communications and Technology 59 18 (March 25, 2021), available at https://www.congress.gov/117/meeting/house/111407/documents/HHRG-117-IF16-Transcript- 19 20210325.pdf; see also Disinformation Nation: Social Media's Role in Promoting Extremism and Misinformation Hearing Before H. Energy and Commerce Subcomm. on Communications and 20 Technology 175 (March 25, 2021), available at https://www.congress.gov/117/meeting/house/111407/documents/HHRG-117-IF16-Transcript- 21 20210325.pdf (Zuckerberg: "[O]ur policies on-on the main apps that we offer generally prohibit people under the age of 13 from using the services."); See also Transcript of Zuckerberg's 22 appearance before House committee, Washington Post (April 11, 2018), https://www.washingtonpost.com/news/the-switch/wp/2018/04/11/transcript-of-zuckerbergs- 23 appearance-before-house-committee/?utm_term=.e7b476fb8ac7&noredirect=on (When asked if it is correct that children can get a Facebook account starting at age 13, Zuckerberg confirmed that 24 it was correct); see also NewSchools Venture Fund, NewSchools Summit 2011: John Doerr and Mark Zuckerberg on innovation and education (May 24, 2011), 25 https://www.youtube.com/watch?v=n03zAOadyMA (Zuckerberg: "[A]nd so basically, we don't allow people under the age of 13 on Facebook . . . today we don't allow people under the age of 26 13 to sign up"). 550 27 Protecting Kids Online: Testimony from a Facebook Whistleblower Hearing before Subcomm. On Consumer Protection, Product Safety, and Data Security (Oct. 5, 2021), available at 28 https://www.commerce.senate.gov/2021/10/protecting%20kids%20online:%20testimony%20fro m%20a%20facebook%20whistleblower. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 131 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 138 of 278 1 Meta is also aware that teenagers coach tweens, defined by them as 10- to 12-year-olds, on how to 2 use its products.551 3 390. Indeed, far from acknowledging the serious defects in its products and warning 4 children and parents of the same, Meta has launched advertising campaigns designed to encourage 5 more children to use its products—by touting the purported safety of those products. For example, 6 in a recent television ad, Meta claimed that it "build[s] technology that gives you more control and 7 helps keep you safe" including through its "industry leading AI" and other "tools that can protect— 8 so you can connect." This advertisement featured children, as in the screenshot below. 9 10 11 12 13 14 15 16 17 18 Other advertising campaigns have similarly touted Meta's AI as being a feature that contributes to 19 its products' safety—without disclosing the serious defects identified in this Complaint. 20 391. In another example of advertising that promotes use by children, a Meta 2021 online 21 advertisement actively highlighted the content available for fifth grade children on its Facebook 22 product, highlighting the experience of an art teacher who used Facebook to communicate with 23 students during the pandemic—an experience the video noted was "a lot to unpack for little, tiny 24 people." 25 391A. In addition, and pursuant to Federal Rule of Civil Procedure 10(c), Plaintiffs adopt 26 by reference the following paragraphs of the Complaint for Injunctive and Other Relief in State of 27 Arizona et al. v. Meta Platforms, Inc., et al., No. 4:23-cv-05448, at ECF No. 73-2, pending in the 28 551 Haugen_00016728 at Haugen_00016736-Haugen_00016740. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 132 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 139 of 278 1 Northern District of California (attached hereto as Exhibit A): 6, 8, 39-43, 54-66, 68-73, 77-85, 2 96, 99-116, 119-150, 160, 162-175, 178-188, 191-219, 228- 254, 256-298, 304-308, 312-372, 379- 3 386, 391-395, 399, 400, 402-429, 435-440, 442, 443, 448-456, 479-497, 499, 528-530, 533-535, 4 539-543, 544- 561, 563- 574, 577- 587, 592, 593, 595-604, 606- 630, 645- 655, 657-677, 679-701, 5 708, 710-724, 726-729, 732- 744, 749, 760, 761, 763-766, 770, 772-791, 796-805, 809, 813-818, 6 820-832. 7 5. Meta facilitates the spread of CSAM and child exploitation. 8 391B. Pursuant to Federal Rule of Civil Procedure 10(c), Plaintiffs also adopt by reference 9 the following paragraphs of the Complaint for Abatement and Civil Penalties in State of New 10 Mexico v. Meta Platforms, Inc., et al., No. D-101-CV-2023-02838, pending in the First Judicial 11 District Court, County of Santa Fe, New Mexico (attached hereto as Exhibit B): 68-72, 74-85, 87- 12 193, 197-199, 201-202, 204-206, 211, 214-215, 217-221, 224-227, 237-242, 246-248, 258, 260- 13 263, 265, and 356-359. 14 392. Various design features of Meta's products promote and dramatically exacerbate 15 sexual exploitation, the spread of CSAM, sextortion, and other socially maladaptive behavior that 16 harms children. 17 393. Meta has long known about these outcomes.552 In 2010, the Daily Mail in the United 18 Kingdom reported that a pedophile used Facebook to groom up to 1,000 children for sex. Detectives 19 "praised the 'brave young people' who helped catch this predator but attacked Meta, saying "many 20 sickening incidents could have been avoided if the social networking site had installed a 'panic 21 button' which allows youngsters to alert authorities if they suspect they were being groomed." 553 22 23 24 25 552 See, e.g., Michael H. Keller and Gabriel J.X. Dance, The Internet Is Overrun With Images of 26 Child Sexual Abuse," N.Y. Times (Sept. 29, 2019), https://www.nytimes.com/interactive/2019/09/28/us/child-sex-abuse.html. 27 553 Michael Seamark, Paedophile postman used Facebook and Bebo to groom up to 1,000 28 children for sex, DailyMail.com (May 28, 2010), https://www.dailymail.co.uk/news/article- 1282157/Facebook-grooming-How-pervert-postman-used-site-groom-hundreds-children.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 133 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 140 of 278 1 394. In 2013, the Christian Science Monitor reported that Facebook is a "favorite 2 recruiting ground[]" for child sex traffickers. 554 3 395. In 2017, The Times in the U.K. reported that Facebook "failed to take down dozens 4 of images and videos that were 'flagged' to its moderators, including . . . several violent paedophilic 5 cartoons" and "a video of an apparent sexual assault on a child." 555 6 396. In 2019, the Sunday Times, also in the U.K., reported that "Instagram is steering 7 paedophiles towards accounts belonging to children as young as 11, who should not be on the 8 platform in the first place."556 9 397. Despite its awareness from over a decade of red flags, Meta promotes its products 10 as safe and family- friendly and claims that its product features are designed to remind adolescent 11 users who they are sharing with and to limit interactions with strangers. 557 This is simply not the 12 case. Meta not only tolerates child exploitation, it knowingly assists, supports, and/or facilitates 13 child exploitation through its defective product features. 14 398. Meta also fails to enforce its own policies regarding adolescent users and does not 15 incorporate simple, cost-effective technologies into the design of its products that would help 16 reduce the prevalence of CSAM. Adolescent users are harmed by Meta's defectively designed 17 products, which are unreasonably dangerous for them. 18 399. For example, Facebook's "People You May Know" feature helps predators connect 19 with underage users and puts them at risk of sexual exploitation, sextortion, and production and 20 distribution of CSAM; 80% of "violating adult/minor connections" on Facebook were the result of 21 22 554 Marjorie Kehe Staff, Kimberly Ritter stands up to child sex trafficking in US hotels, The 23 Christian Science Monitor (Mar. 15, 2013), https://www.csmonitor.com/World/Making-a- difference/2013/0315/Kimberly-Ritter-stands-up-to-child-sex-trafficking-in-US-hotels. 24 555 Alexi Mostrous, Facebook publishing child pornography, The Times (Apr. 13, 2017), https://www.thetimes.co.uk/article/facebook-publishing-child-pornography- 25 pdgt87nm6?region=global. 556 Shanti Das & Geoff White, Instagram sends paedophiles to accounts of children as young as 26 11, The Sunday Times (Dec. 1, 2019), https://www.thetimes.co.uk/article/instagram-sends- predators-to-accounts-of-children-as-young-as-11-j2gn5hq83. Meta was aware of this report. 27 META3047MDL-003-00153063. 557 28 Safety Resources for Parents, Meta Privacy, Safety, and Security https://www.facebook.com/help/1079477105456277?helpref=faq_content. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 134 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 141 of 278 1 this friends recommendation system.558 Instagram's "Suggested for You" and "Because You 2 Watched" features are similarly dangerous because they connect strangers, including adult 3 predators, with adolescent users. As The Sunday Times revealed, "[p]redators who follow users 4 posting photos of young models, dancers or gymnasts are shown a stream of other images they will 5 like and targeted with personalised recommendations of accounts to follow. Among the suggested 6 accounts are newly created profiles belonging to children who would otherwise be almost 7 impossible to find unless you had their user name." 559 8 400. Similarly, the absence of effective age verification measures, as described above, 9 allows predators to lie about their ages and masquerade as children, with obvious dangers to the 10 actual children on Meta's products. Prior to November 2022, the default setting for Facebook users' 11 profiles allowed posts to be publicly viewable by any user. This allowed predators to discover and 12 connect with adolescent users. The same is true for users' friends lists. 13 401. Instagram is similarly flawed, having transitioned to private profiles for users under 14 16 only in July 2021. Up until that change—and even after—millions of minors are left exposed to 15 predation and at risk of extortion and abuse by default. Indeed, The Sunday Times reported that 16 "[o]ne of those brought to the surface by Instagram's algorithm contained selfies of a young girl 17 and a profile description that read: "Hey people hope you decide to follow me im 11." 560 18 402. Distressingly, Meta considered making teenage users' profiles "private by default" 19 at least as early as July 2020, but chose not to do so after pitting "safety, privacy, and policy wins" 20 against "growth impact."561 21 403. Meta's products also include direct messaging features. Instagram's direct 22 messaging system is equipped with a product feature called a "photo bomb," which is an image or 23 558 24 META3047MDL-003-00013254 at META3047MDL-003-00013255. 559 Shanti Das & Geoff White, Instagram sends paedophiles to accounts of children as young as 25 11, The Sunday Times (Dec. 1, 2019), https://www.thetimes.co.uk/article/instagram-sends- predators-to-accounts-of-children-as-young-as-11-j2gn5hq83. 26 560 Shanti Das & Geoff White, Instagram sends paedophiles to accounts of children as young as 11, The Sunday Times (Dec. 1, 2019), https://www.thetimes.co.uk/article/instagram-sends- 27 predators-to-accounts-of-children-as-young-as-11-j2gn5hq83. 561 28 META3047MDL-003-00028226 at META3047MDL-003-00028226; META3047MDL-003- 00013254 at META3047MDL-003-00013254. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 135 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 142 of 278 1 video sent from a smartphone that automatically disappears from the recipient's inbox. Both 2 Facebook's and Instagram's messaging system also have a "Vanish Mode" option, which makes 3 the message disappear after it has been read. 4 404. Meta's messaging features allow users to exchange private messages with other 5 product users. In addition, users do not have to be connected as friends or followers to initiate 6 conversations, which enables predators to communicate privately with youth, with virtually no 7 evidence of what was exchanged. This feature enables predators to identify children who are willing 8 to respond to a stranger's message, and then prey on their insecurities. Even though "this is the kind 9 of thing that pisses Apple off to the extent of threatening to remove us from the App Store," as of 10 mid-2020, Meta had no timeline for "when we'll stop adults from messaging minors in IG 11 Direct."562 That remained true even after Meta received reports that a 12-year-old minor solicited 12 on its platform "was [the] daughter of [an] Apple Security Exec." 563 13 405. An internal study conducted in or around June of 2020 concluded that 500,000 14 underage Instagram accounts "receive IIC"—which stands for "inappropriate interactions with 15 children"—on a daily basis.564 Yet, at the time, "Child Safety [was] explicitly called out as a non- 16 goal . . . . So if we do something here, cool. But if we can do nothing at all, that's fine, too." 565 17 406. Meta's products also permit users to operate multiple accounts simultaneously. 18 Operating multiple accounts enables adolescent users to have multiple unique online identities. In 19 addition, parents are often unaware that more than one account exists and therefore do not monitor 20 the additional accounts as they would the primary, known account. By permitting multiple 21 accounts, Meta compounds children's exposure to danger on its products and hampers parents' 22 attempts to monitor their children's activities. 23 407. Meta's products also utilize a location feature that allows users to geotag the location 24 where a photo was taken or from where a post is being made. On Facebook, users can search posts 25 562 26 META3047MDL-003-00028019 at META3047MDL-003-00028019. 563 META3047MDL-003-00028019 at META3047MDL-003-00028020. 27 564 META3047MDL-003-00028214 at META3047MDL-003-00028216- META3047MDL-003- 00028218. 28 565 META3047MDL-003-00028214 at META3047MDL-003-00028215. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 136 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 143 of 278 1 by location and find pages and groups by the location tagged in a user's post. Similarly, Instagram 2 users can use the Explore tool to search for posts based on location tags. 3 408. Location tagging is inherently dangerous for children, as it identifies where they are 4 located, where they vacation, where they attend school, and so on. Predators can find these posts 5 by searching within specific geographic confines. This enables the identification of potential 6 victims in a predator's area, increasing the risk that adolescent users are targeted for sexual 7 exploitation, sextortion, and CSAM. 8 409. Meta's policies fail to adequately protect children, especially teens. Meta created its 9 own definition of CSAM that fails to sufficiently meet the clear requirements provided in 18 U.S.C. 10 § 2256(8) and related case law. Meta relies on its own definitions to fail to report harmful CSAM 11 to the authorities as required by law.566 For example, Meta utilizes the Tanner Stages, a 12 classification system used to track children's physical development during puberty, to assist with 13 making moderation decisions related to potential CSAM. The scale's creator, Dr. James Tanner, 14 has called this approach "wholly illegitimate."567 15 410. Despite using PhotoDNA and other technology in Facebook's product design as 16 early as 2011, Meta has hindered its effectiveness and success by creating its own CSAM 17 definitions and compromising its own detection model. 18 411. In fact, the United States Department of Justice ("DOJ") urged Zuckerberg to refrain 19 from implementing dangerous design modifications to his products, "embed the safety of the public 20 in system designs," and "act against illegal content effectively with no reduction to safety," in ways 21 that safeguard victims.568 22 412. In November of 2021, Meta indicated that it would postpone certain product design 23 changes, such as encrypting direct messages on Instagram, that would create an increased risk and 24 566 Michael H. Keller, Adults or Sexually Abused Minors? Getting It Right Vexes Facebook, N.Y. 25 Times (Mar. 31, 2022), https://www.nytimes.com/2022/03/31/business/meta-child-sexual- abuse.html. 26 567 Michael H. Keller, Adults or Sexually Abused Minors? Getting It Right Vexes Facebook, New York Times, (March 31, 2022), https://www.nytimes.com/2022/03/31/business/meta-child- 27 sexual-abuse.html. 568 28 Letter to Mark Zuckerberg from Department of Justice 2 (October 4, 2019), https://www.justice.gov/opa/press-release/file/1207081/download . PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 137 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 144 of 278 1 volume of CSAM within its products. However, in January 2022, it implemented those changes to 2 its Messenger application, increasing risks to vulnerable children and the volume of predators and 3 CSAM without sufficient warning.569 In 2019, FBI Director Christopher Wray stated that, with the 4 design decision to encrypt Messenger absent additional protections for children, Facebook would 5 become "a dream-come-true for predators and child pornographers. A platform that allows them to 6 find and connect with kids, and like-minded criminals, with little fear of consequences. A lawless 7 space created not by the American people, or their elected officials, but by the owners of one big 8 company."570 9 413. Even further compounding these problems, Meta has "instructed content moderators 10 for its platforms to 'err on the side of an adult' when they are uncertain about the age of a person 11 in a photo or video, according to a corporate training document." 571 12 414. Shortly after Frances Haugen disclosed how Meta's products harm children, an 13 unnamed whistleblower and former Facebook employee revealed in a five-page document that 14 Meta's efforts to address the prevalence of CSAM within its products were "inadequate" and 15 "under-resourced."572 This whistleblower also stated that Meta "doesn't track" the full scale of the 16 CSAM problem within its products because senior executives consistently limit the funds available 17 for child protection design efforts, by instead focusing on the company's "return on investment." 573 18 415. Meta's failure to monitor its products for CSAM and protect its most vulnerable 19 users is all the more shocking considering the troves of data and information it collects about users 20 to monitor their preferences and lifestyles for advertising clients. Using that same technology, Meta 21 569 Timothy Buck, Express Yourself in Messenger's End-to-End Encrypted Chats, Messenger 22 News (Jan. 27, 2022), https://messengernews.fb.com/2022/01/27/express-yourself-in-messengers- end-to-end-encrypted-chats/. 570 23 Raphael Satter & Sarah N. Lynch, FBI Director Warns Facebook Could Become Platform Of 'Child Pornographer', Reuters (Oct. 4, 2019), https://www.reuters.com/article/us-facebook- 24 security/fbi-director-warns-facebook-could-become-platform-of-child-pornographers- idUSKBN1WJ1NQ. 25 571 Michael H. Keller, Adults or Sexually Abused Minors? Getting It Right Vexes Facebook, N.Y. Times (Mar. 31, 2022), https://www.nytimes.com/2022/03/31/business/meta-child-sexual- 26 abuse.html. 572 Angus Crawford, Whistleblower: Facebook's response to child abuse 'inadequate', BBC 27 News, (Oct. 28, 2021) https://www.bbc.com/news/technology-59063768. 573 28 Angus Crawford, Whistleblower: Facebook's response to child abuse 'inadequate', BBC News, (Oct. 28, 2021) https://www.bbc.com/news/technology-59063768. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 138 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 145 of 278 1 could easily detect, report and take actions to prevent instances of sexual grooming, sextortion, and 2 CSAM distribution on its products. 3 416. Instead of taking these crucial, life-saving actions, Meta misleadingly asserts that 4 when it "become[s] aware of apparent child exploitation, we report it to the National Center for 5 Missing and Exploited Children (NCMEC), in compliance with applicable law." 574 But Meta's 6 response to law enforcement inquiries is often significantly delayed, if they respond at all, and Meta 7 further frustrates law enforcement investigations by failing to promptly to report child sexual 8 exploitation.575 9 417. As a result, Meta's products are polluted with illegal material that promotes and 10 facilitates the sexual exploitation of minors. Meta benefits from increased user activity (and 11 increased advertising revenue) for disseminating these materials on its products. 12 418. Meta knows that its products are used for the production, possession, distribution, 13 receipt, transportation, and dissemination of millions of materials that depict obscene visual 14 representations of the sexual abuse of children each year. Meta also knows that its defective 15 algorithms worsen the problem: "CEI (Child Expolitative [sic] Imagery) is . . . something people 16 seek out, and our recommendations can make worse." 576 17 419. Meta knowingly fails to take adequate and readily available measures to remove 18 these contraband materials from its products in a timely fashion. 19 420. In violation of 18 U.S.C. § 2258A, Meta knowingly fails to report massive amounts 20 of material in violation of 18 U.S.C. § 2256 and 18 U.S.C. § 1466A. 21 22 23 24 574 Meta, Meta's Child Sexual Exploitation, Abuse and Nudity, Facebook Community Standards, https://transparency.fb.com/policies/community-standards/child-sexual-exploitation-abuse- 25 nudity/. 575 See Michael H. Keller & Gabriel J. X Dance, The Internet Is Overrun With Images Of Child 26 Sexual Abuse. What Went Wrong?, N.Y. Times (Sept. 29, 2019), https://www.nytimes.com/interactive/2019/09/28/us/child-sex-abuse.html (describing the 27 proliferation of CSAM on social media apps and the way the apps have hampered law enforcement investigations). 28 576 META3047MDL-003-00068860 at META3047MDL-003-00068861. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 139 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 146 of 278 1 421. Meta knows, or reasonably should have known, that its products are increasingly 2 unsafe for children each year, and yet fails to implement safeguards to prevent children from 3 accessing its products. 4 422. In addition to these dangerous features that enable CSAM and child exploitation, 5 Meta's products hamper identification and reporting of CSAM or child pornography. For example, 6 they do not allow a person to report harmful content without first logging into and using the 7 products, which requires them to sign up for an account and provide a name and email address. 577 8 423. Neither Instagram nor Facebook contain product features that allow users to report 9 harmful images or videos directly from their direct messaging features. 578 10 424. Upon information and belief, Meta paused or completely stopped certain proactive 11 scanning measures related to child exploitation imagery and CSAM for some unknown period(s), 12 including a period within the past three years. 579 13 425. Meta knowingly possessed the capabilities and technologies to incorporate other 14 automatic actions into its product designs to protect children (including but not limited to 15 immediately disabling or deleting harmful content to minors) but Meta deliberately and willfully 16 failed to do so. Instead, Meta allowed the sexualization and exploitation of minors on their products 17 to brazenly continue. 18 426. Meta knowingly failed to invest in adequate CSAM prevention measures, including 19 but not limited to client-side scanning and perceptual hashing. 20 427. Despite having the technology to limit the spread, Meta continues to fail to prevent 21 the spread of this same CSAM on their Products. 580 22 577 23 Canadian Centre for Child Protection, Reviewing Child Sexual Abuse Material Reporting Functions on Popular Platforms 16, 24 https://protectchildren.ca/pdfs/C3P_ReviewingCSAMMaterialReporting_en.pdf (last accessed December 28, 2022). 25 578 Canadian Centre for Child Protection, Reviewing Child Sexual Abuse Material Reporting Functions on Popular Platforms 13, 26 https://protectchildren.ca/pdfs/C3P_ReviewingCSAMMaterialReporting_en.pdf (last accessed December 28, 2022). 27 579 META3047MDL-003-00009133 at META3047MDL-003-00009134. 580 28 See META3047MDL-003-00012994 at META3047MDL-003-00012995- META3047MDL- 003-00012996 (describing Meta's adoption of different CSAM prevention technologies). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 140 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 147 of 278 1 428. Meta knowingly failed to design its products to proactively detect harmful 2 interactions between adults and minors despite having the technology and capabilities to do so 3 successfully.581 4 429. Finally, Meta's products offer unreasonably inadequate parental controls; for 5 example, parents cannot monitor their child's account without logging into the child's account 6 directly. 7 430. Collectively, these defects make it difficult for parents to monitor their use of Meta's 8 products, and they enable predators to identify, connect to, and exploit children. 582 9 6. Meta failed to adequately warn Plaintiffs or Consortium Plaintiffs about the dangers and harms caused by Instagram and Facebook, or 10 provide instructions regarding safe use. 11 431. Meta has failed to adequately warn adolescent users and parents about the physical 12 and mental health risks posed by Instagram and Facebook. These risks include a plethora of mental 13 health disorders like compulsive use, addiction, eating disorders, anxiety, depression, insomnia, 14 exacerbated executive dysfunction, sexual exploitation from adult users, suicidal ideation, self- 15 harm, and death. 16 432. Meta targets adolescent users via advertising and marketing materials distributed 17 throughout digital and traditional media that fail to provide sufficient warnings to potential 18 adolescent consumers of the physical and mental risks associated with using Facebook and 19 Instagram. 20 433. Meta also fails to adequately warn adolescent users during the product registration 21 process. At account setup, neither Instagram nor Facebook contain warning labels, banners, or 22 conspicuous messaging to adequately inform adolescent users of the known product risks and 23 potential physical and mental harms associated with usage. Instead, Meta allows adolescent users, 24 25 581 Hany Farid, Reining in Online Abuses, 19 Technology and Innovation 593-599 (2018); 26 META3047MDL-003-00009133 at META3047MDL-003-00009134 (describing Meta's pause of certain CSAM prevention work during Covid-19 and CSAM prevention procedures more 27 broadly). 582 28 Hany Farid, Reining in Online Abuses, 19 Technology and Innovation 593-599 (2018), https://farid.berkeley.edu/downloads/publications/nai18.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 141 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 148 of 278 1 including those under the age of 13, to easily create an account (or multiple accounts) and fully 2 access these products. 3 434. Meta's failure to warn adolescent users continues even as adolescents exhibit 4 problematic signs of addiction to and compulsive use of Facebook or Instagram. For example, Meta 5 does not warn users when their screen time reaches harmful levels or when adolescents are 6 accessing the product habitually. 7 435. Despite proactively providing adolescent users with countless filtering and editing 8 tools, Meta also does not appropriately warn adolescent users regarding which images have been 9 altered or the mental health harms associated with the heavily filtered images that Meta presents 10 and recommends. 11 436. Not only does Meta fail to adequately warn users regarding the risks associated with 12 Instagram and Facebook, it also does not provide sufficient instructions on how adolescents can 13 safely use the products. 14 437. Meta's failure to adequately warn and instruct as set forth herein has proximately 15 caused significant harm to the mental and physical well-being of Plaintiffs and Consortium 16 Plaintiffs, in addition to the other injuries and harms as set forth herein. 17 C. FACTUAL ALLEGATIONS AS TO SNAP 18 438. Snap Inc. calls itself "a camera company." 583 Its "flagship product, Snapchat, is a 19 camera application that was created to help people communicate through short videos and images. 20 [Snap] calls each of those short videos or images a Snap." 584 Snap's design of its Snapchat product 21 capitalizes on children's increasing attachment to quick, instantaneous exchanges. As Snap's 22 founder and CEO Evan Spiegel has explained, "today… pictures are being used for talking. So 23 when you see your children taking a zillion photos of things that you would never take a picture of, 24 583 Snap Inc. Form S-1 Registration Statement (hereafter "Form S-1") at 1 (Feb. 2, 2017), 25 https://www.sec.gov/Archives/edgar/data/1564408/000119312517029199/d270216ds1.htm; See also, Snap – Who We Are, Snap Inc.; ("We believe that reinventing the camera represents our 26 greatest opportunity to improve the way people live and communicate.").https://careers.snap.com/en-US ("We believe that reinventing the camera 27 represents our greatest opportunity to improve the way people live and communicate."). 584 28 Snap Inc. Form S-1 Registration Statement (hereafter "Form S-1") at 1 (Feb. 2, 2017), https://www.sec.gov/Archives/edgar/data/1564408/000119312517029199/d270216ds1.htm. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 142 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 149 of 278 1 it's cos [sic] they're using photographs to talk. And that's why people are taking and sending so 2 many pictures on Snapchat every day."585 3 439. Spiegel's statement is telling, as Snap has tailored every aspect of its Snapchat 4 product to children rather than adults. Snap designed and implemented dangerous features in 5 Snapchat that exploit children's need for social acceptance and rewards by pushing its users to 6 maximize their use of and engagement with the app. Snap built Snapchat using manipulative 7 techniques to compel young users to send an ever-increasing number of photographs and videos, 8 and to reward users who maximize their engagement with elevated status. Snap also dangerously 9 encourages adolescents to increase engagement on the app indiscriminately, pushing tools to share 10 sensitive material with an ever-expanding group of friends and strangers. 11 440. Snapchat's design features cause its young users to suffer increased anxiety, 12 depression, disordered eating, sleep deprivation, suicide, and other severe mental and physical 13 injuries. Snap knows or should have known this. Snap intentionally designed Snapchat to prey on 14 the neuropsychology and behavioral patterns of children to maximize their engagement and 15 increase Snap's advertising revenue. Despite this knowledge, Snap continues to update its product 16 and add features intentionally designed to entice, exploit, and addict kids, including Snap Streaks, 17 trophies, social signifiers and reward systems, quickly disappearing messages, filters, lenses, and 18 games. 19 441. Snap knew, or should have known, that its conduct has negatively affected youth. 20 Snap's conduct has been the subject of inquiries by the United States Senate regarding Snapchat's 21 use "to promote bullying, worsen eating disorders, and help teenagers buy dangerous drugs or 22 engage in reckless behavior."586 Further, Senators from across the ideological spectrum have 23 24 25 585 Stuart Dredge, What is Snapchat? CEO Evan Spiegel explains it all for parents, The Guardian, June 15, 2015, https://www.theguardian.com/technology/2015/jun/17/what-is-snapchat-evan- 26 spiegel-parents. 586 27 Bobby Allyn, 4 Takeaways from the Senate child safety hearing with YouTube, Snapchat and TikTok, National Public Radio (Oct. 26, 2021), 28 https://www.npr.org/2021/10/26/1049267501/snapchat-tiktok-youtube-congress-child-safety- hearing. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 143 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 150 of 278 1 introduced bills that would ban many of Snapchat's features that are particularly addictive to 2 adolescents.587 3 442. Despite these calls for oversight from Congress, Snap has failed to curtail its use of 4 features such as streaks, badges, and other awards that reward users' level of engagement with 5 Snapchat. As described in detail below, Snapchat is a product that causes harm to children, the 6 target audience for whom Snap designed and to whom it promoted its product. 7 1. Background and overview of Snapchat. 8 443. Snapchat was created by three college students in 2011 and first released for iPhones 9 in September 2011. Snapchat quickly evolved from its origin as a disappearing-message chat 10 application after Snap's leadership made design changes and rapidly developed new product 11 features. As a result of its design and implementation of dangerous and addictive features 12 specifically targeting youths (described below), Snapchat quickly became widely used among 13 children. 14 444. Snap marketed Snapchat as "temporary social media" that would allow users to 15 show a more authentic, unpolished, and spontaneous side of themselves. 588 Snapchat's central and 16 defining feature, the "Snap," allows users to send and receive ephemeral, or "disappearing," 17 audiovisual messages. That feature foreseeably and quickly drove users to exchange sexually 18 explicit "Snaps," sometimes called "sexts" even though they are photos. Because of its brand 19 identity among millennials as the original ephemeral-messaging app, Snapchat almost immediately 20 became known as the "sexting" app—a fact that Snap was or should have been on notice of from 21 public sources.589 22 587 See Abigal Clukey, Lawmaker Aims To Curb Social Media Addiction With New Bill, National 23 Public Radio (Aug. 3, 2019), https://www.npr.org/2019/08/03/747086462/lawmaker-aims-to- curb-social-media-addiction-with-new-bill; Social Media Addiction Reduction Technology Act, 24 S. 2314, 116th Cong. (2019); Kids Internet Design and Safety Act, S. 2918, 117th Cong. (2021). 588 Jenna Wortham, A Growing App Lets You See It, Then You Don't, New York Times (Feb. 9, 25 2013), https://www.nytimes.com/2013/02/09/technology/snapchat-a-growing-app-lets-you-see-it- then-you-dont.html?_r=0. 26 589 Megan Dickey, Let's Be Real: Snapchat Is Totally Used For Sexting, Bus. Insider (Nov. 30, 27 2012), https://www.businessinsider.com/snapchat-growth-sexting-2012-11; Billy Gallagher, No, Snapchat Isn't About Sexting, Says Co-Founder Evan Spiegel, TechCrunch (May 12, 2012), 28 https://techcrunch.com/2012/05/12/snapchat-not-sexting/b (describing an interview in which a journalist asked the CEO of Snap about the product's potential use for sexting). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 144 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 151 of 278 1 445. Snapchat creates images and GIFs for users to incorporate into their videos and 2 picture postings. Snap has also acquired publishing rights to thousands of hours of music and video 3 which it provides to Snapchat users to attach to the videos and pictures that they send. 4 2. Snap targets children. 5 a. Snap has designed its Snapchat product to grow use by children to drive the company's revenue. 6 7 446. Within five months of launching, Snapchat had 40,000 users. 590 By May 2012, less 8 than eight months after launching, CEO Evan Spiegel reported that the company was "thrilled" to 9 learn that most of Snapchat's users were high school students sending "behind-the-back photos of 10 teachers and funny faces" to each other during class. According to Spiegel, Snap's server data 11 showed peaks of activity during the school day. 591 12 447. Snap immediately focused on increasing the product's frequency of use. 592 By late 13 2012, Snapchat had over a million active users sending over 20 million Snaps per day. 593 By 2013, 14 Snapchat users were sending over 60 million Snaps per day. 594 By the end of 2022, this number has 15 risen to over 5 billion Snaps per day. 595 16 448. As Snap continued to quickly add new features to its product, the number of 17 Snapchat's daily active users (users who open Snapchat at least once during a 24-hour period) 18 19 20 590 Ken Auletta, Get Rich U, New Yorker (Apr. 30, 2012), 21 https://www.newyorker.com/magazine/2012/04/30/get-rich-u. 591 22 Team Snapchat, Let's Chat, Snapchat Blog at http://blog.snapchat.com (May 9, 2012), available at https://web.archive.org/web/20120518003029/http://blog.snapchat.com:80/. 592 23 Billy Gallagher, You Know What's Cool? A Billion Snapchats: App Sees Over 20 Million Photos Shared Per Day, Releases On Android, TechCrunch (Oct. 29, 2012), 24 https://techcrunch.com/2012/10/29/billion-snapchats/. 593 Billy Gallagher, You Know What's Cool? A Billion Snapchats: App Sees Over 20 Million 25 Photos Shared Per Day, Releases On Android, TechCrunch (Oct. 29, 2012), https://techcrunch.com/2012/10/29/billion-snapchats/. 26 594 Billy Gallagher, Snapchat Raises $13.5M Series A Led By Benchmark, Now Sees 60M Snaps Sent Per Day, TechCrunch (Feb. 9, 2013), https://techcrunch.com/2013/02/08/snapchat-raises-13- 27 5m-series-a-led-by-benchmark-now-sees-60m-snaps-sent-per-day/. 595 28 Snap Inc. Q4 2022 Investors Meeting Transcript at p. 7 (Jan. 31, 2023), https://s25.q4cdn.com/442043304/files/transcript/snap-inc.-q4-2022-transcript.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 145 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 152 of 278 1 rapidly increased.596 In 2017, Snap reported that its users opened the product more than 18 times a 2 day on average. By 2019, users were opening the product an average of 30 times per day. 3 4 5 6 7 8 9 10 11 12 13 597 14 449. Today, Snapchat is one of the world's most widely used apps. By its own estimates, 15 Snapchat has 363 million daily users, including 100 million daily users in North America. 598 16 Snapchat also "reaches 90% of the 13-24 year old population" in over twenty countries, and reaches 17 nearly half of all smartphone users in the United States. 599 18 450. Snapchat's explosive growth is driven by its key user demographic, 13-17 year olds. 19 In 2022, 59% of US teens used Snapchat and 15% said they used it "almost constantly." 600 Snapchat 20 proudly touts its influence over what it calls the "Snapchat Generation" ("Gen Z"). 601 21 22 596 Snap Inc. Form S-1 Registration Statement (hereafter "Form S-1") at 91 (Feb. 2, 2017), 23 https://www.sec.gov/Archives/edgar/data/1564408/000119312517029199/d270216ds1.htm. 597 24 Snap Inc. Form S-1 Registration Statement (hereafter "Form S-1") at 91 (Feb. 2, 2017), https://www.sec.gov/Archives/edgar/data/1564408/000119312517029199/d270216ds1.htm. 598 25 October 2022 Investor Presentation at 5, Snap Inc. (Oct. 20, 2022), https://investor.snap.com/events-and-presentations/presentations/default.aspx. 26 599 October 2022 Investor Presentation at 6-7, Snap Inc. (Oct. 20, 2022), https://investor.snap.com/events-and-presentations/presentations/default.aspx. 27 600 Pew Research Center, Teens, Social Media and Technology 2022 (Aug. 10, 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology-2022/. 28 601 SNAP0000137 at 0139. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 146 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 153 of 278 1 451. In 2014, Snap began running advertisements on Snapchat. 602 Snapchat's entire 2 business model revolves around its advertising revenue. According to internal company records, 3 advertisements were pervasive on Snapchat by 2015 and, by 2018, 99% of Snap's total revenue 4 came from advertising. Advertising has accounted for 99% of Snap's revenue each year since 5 2018.603 In 2022, Snap's revenue was approximately $4.6 billion.604 6 452. Snap attracts advertisers by providing them access to the huge universe of Snapchat 7 users and by collecting immense amounts of data on its users, including its pre-teen and teenage 8 users, which it uses to target advertising to those users. Snap makes no secret of this practice, 9 recently acknowledging that it relies "heavily on our ability to collect and disclose data, and metrics 10 to our advertisers so we can attract new advertisers and retain existing advertisers. Any restriction 11 or inability, whether by law, regulation, policy, or other reason, to collect and disclose data and 12 metrics which our advertisers find useful would impede our ability to attract and retain 13 advertisers."605 14 453. Snap's growth in advertising revenues was driven by changes Snap made to 15 Snapchat that incentivized compulsive and addictive use at the expense of its users' health. Snap's 16 internal research indicates the Snapchat experience is "more immersive" than its competitors' apps. 17 This means users are more likely than on other apps to keep watching videos (and advertising). 606 18 Other research shows that Snapchat's daily active users are constantly using its product; compared 19 to other apps, users are most likely to use Snapchat "right when I wake up," "before work/school," 20 "during work/school," "after work/school," "on vacations," and "when I'm with others[.]" 607 21 454. Snap understands that its user experience must be immersive and all-encompassing 22 in order to maximize its advertising revenue. Indeed, Snap recently admitted to its investors that its 23 602 Angela Moscaritolo, Snapchat Adds 'Geofilters' in LA, New York, PC Mag. (July 15, 2014), https://www.pcmag.com/news/snapchat-adds-geofilters-in-la-new-york. 24 603 Snap Inc. Form 10-K at 18 (Dec. 31, 2022), https://d18rn0p25nwr6d.cloudfront.net/CIK- 25 0001564408/c22ae9bd-7418-456e-82d4-48129de1df54.pdf. 604 Snap Inc. Form 10-K at 18 (Dec. 31, 2022), https://d18rn0p25nwr6d.cloudfront.net/CIK- 26 0001564408/c22ae9bd-7418-456e-82d4-48129de1df54.pdf. 605 Snap Inc. Form 10-K at 18 (Dec. 31, 2022), https://d18rn0p25nwr6d.cloudfront.net/CIK- 27 0001564408/c22ae9bd-7418-456e-82d4-48129de1df54.pdf. 606 SNAP0000103 at 0120. 28 607 SNAP0000103 at 0113. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 147 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 154 of 278 1 revenue could be harmed by, among other things, "a decrease in the amount of time spent on 2 Snapchat, a decrease in the amount of content that our users share, or decreases in usage of our 3 Camera, Visual Messaging, Map, Stories, and Spotlight platforms." 608 4 b. Snap promotes Snapchat to children. 5 455. Snap specifically promotes Snapchat to children because they are a key 6 demographic for Snap's advertising business. 7 456. In its first post on its website, Snapchat observed that "[t]o get a better sense of how 8 people were using Snapchat and what we could do to make it better, we reached out to some of our 9 users. We were thrilled to hear that most of them were high school students who were using 10 Snapchat as a new way to pass notes in class—behind-the-back photos of teachers and funny faces 11 were sent back and forth throughout the day." 609 12 457. As shown in this capture of a Snapchat feature page created by Snap, Snap uses 13 bright colors, cartoonish designs, and other features that appeal to younger audiences. 14 15 16 17 18 19 20 21 22 23 24 25 26 608 Snap Inc. Form 10-K at 19 (Dec. 31, 2022), https://d18rn0p25nwr6d.cloudfront.net/CIK- 27 0001564408/c22ae9bd-7418-456e-82d4-48129de1df54.pdf. 609 28 Team Snapchat, Let's Chat, Snapchat Blog at http://blog.snapchat.com (May 9, 2012), https://web.archive.org/web/20120518003029/http://blog.snapchat.com:80/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 148 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 155 of 278 1 458. Similarly, in an October 2019 interview, Snap's CEO explained that "we've seen a 2 lot of engagement with our 13-34 demographic, which for us is strategically a critical demographic, 3 not only because that's a demographic that enjoys using new products but also because I think they 4 represent, really, the future . . . So that's obviously been a group that's been really fun to build for, 5 and really it started because those are our friends."610 6 459. Snap touts to advertisers its ability to use Snapchat to reach children. In a December 7 2022 statement to advertisers, Snap claimed that "Snapchat delivers on the emotions that Gen Z 8 seeks and it does so consistently across the platform in areas like Discover, Stories and the 9 Camera."611 To prove that, Snapchat "used a neuroscience measurement called Immersion to 10 measure reactions to different brand messaging—specifically brand purpose messaging vs. non- 11 brand purpose messaging. Immersion captures attention and emotional resonance through 12 variations in heart rate rhythm collected by smartwatches." 612 Per Snapchat, "[a]ny brand or 13 marketer can get on any app and start targeting Gen Z [emphasis added]. After all, Gen Z is 14 digitally native. But to effectively connect and engage with this generation, that takes a different, 15 more intentional type of platform- Snapchat." 613 16 460. Advertisers have responded, pouring into Snapchat money clearly intended for 17 advertising aimed at children. Brands like candy manufacturer Sour Patch Kids, children's toy store 18 ToysRUs, and sugary beverage seller Kool-Aid have all run successful advertising campaigns 19 through Snapchat, frequently using augmented reality tools developed in collaboration with 20 Snapchat. 21 22 23 24 610 Evan Spiegel, Co-Founder and CEO of Snap, Inc., Goldman Sachs, at 4:43-6:23. (Oct. 2, 25 2019), https://www.youtube.com/watch?v=PQiKv-GCQ-w. 611 Snap for Business, What Does Gen Z Want From Brands? Dec. 15, 2022), 26 https://forbusiness.snapchat.com/en-US/blog/what-does-gen-z-want. 612 Snap for Business, What Does Gen Z Want From Brands? Dec. 15, 2022), 27 https://forbusiness.snapchat.com/en-US/blog/what-does-gen-z-want 613 28 Snap for Business, What Does Gen Z Want From Brands? Dec. 15, 2022), https://forbusiness.snapchat.com/en-US/blog/what-does-gen-z-want. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 149 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 156 of 278 1 2 3 4 5 6 7 8 9 10 11 12 13 14 461. Snapchat's age verification systems are defective. For the first two years of its 15 existence, Snap did not even purport to limit user access to those 13 or older. 614 Users were not 16 required to input a date of birth when creating an account. 615 17 462. In 2013, Snap belatedly introduced age limits (which, as explained below, it does 18 not effectively enforce). At the same time, Snap launched a new feature called "Snapkidz" aimed 19 at and designed to attract younger children users, while hedging against the potential user loss due 20 to the new age limits. The Snapkidz feature allowed children under the age of 13 to take filtered 21 photos, draw on them, save them locally on their devices, send them to others, and upload them to 22 other apps.616 Although this version prevented children from sharing "Snaps" on the product, it 23 nonetheless exposed children to Snapchat's features, which normalized and acclimatized children 24 to using Snapchat. In addition, nothing prevented children from creating an unrestricted account 25 614 Team Snapchat, iOS Update: Bug Fixes and More!, Snapchat Blog (June 22, 2013), 26 https://web.archive.org/web/20130627073951/http://blog.snapchat.com:80/. 615 Team Snapchat, iOS Update: Bug Fixes and More!, Snapchat Blog (June 22, 2013), 27 https://web.archive.org/web/20130627073951/http://blog.snapchat.com:80/. 616 28 Team Snapchat, iOS Update: Bug Fixes and More!, Snapchat Blog (June 22, 2013), https://web.archive.org/web/20130627073951/http://blog.snapchat.com:80/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 150 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 157 of 278 1 with a false date of birth on Snapchat and using the product outside the SnapKidz's limited 2 features.617 3 463. The SnapKidz feature was discontinued in or around 2016. Snap now purports to 4 prohibit users under the age of 13. But nothing prohibits the minor user from simply altering their 5 birthdate during the same session where they were just denied an account for being an underage 6 user. Snap could have implemented robust, effective age verification protocols. Instead, it has set 7 up its business and product so that nothing is done to verify the age of its users or to enforce its age 8 limitations. Snap could, but intentionally does not, verify the phone number, email address, or 9 birthdate used to create accounts, and it allows users to create multiple accounts using the same 10 email address or phone number. 11 464. Snap's executives have admitted that Snapchat's age verification "is effectively 12 useless in stopping underage users from signing up to the Snapchat app." 618 Not surprisingly, 13 underage use is widespread. As of 2021, 13% of children ages 8-12 use Snapchat. 619 14 465. Once Snapchat is installed on a user's mobile phone, the product continues to 15 download and install updates, design changes, and new features from Snapchat directly to its users. 16 466. Similarly, the absence of effective age-verification measures means that users who 17 are older can claim to be children—which is an obvious danger to the actual children on Snap's 18 product. 19 20 21 22 617 See Larry Magid, Snapchat Creates SnapKidz – A Sandbox for Kids Under 13, Forbes (June 23 23, 2013), https://www.forbes.com/sites/larrymagid/2013/06/23/snapchat-creates-snapkidz-a- sandbox-for-kids-under-13/?sh=7c682a555e5a; Anthony Cuthbertson, Snapchat admits its age 24 verification system does not work, Independent (Mar. 19, 2019), https://www.independent.co.uk/tech/snapchat-age-verification-not-work-underage-ageid- 25 a8829751.html. 618 Isobel Asher Hamilton, Snapchat admits its age verification safeguards are effectively useless, 26 Bus. Insider (Mar. 19, 2019), https://www.businessinsider.com/snapchat-says-its-age-verification- safeguards-are-effectively-useless-2019-3. 27 619 Victoria Rideout et al., Common Sense Census: Media use by tweens and teens, 2021 at 5, 28 Common Sense Media, https://www.commonsensemedia.org/sites/default/files/research/report/8- 18-census-integrated-report-final-web_0.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 151 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 158 of 278 1 3. Snapchat is designed to addict children through psychological manipulation. 2 3 467. Once Snap entices children to use its product, it uses a series of product features that 4 are designed to addict children. As laid out below, those features can be broadly grouped into two 5 categories that exploit techniques discussed earlier in this Complaint. The first are social metrics 6 and other similar psychological manipulation techniques. The second are features designed to 7 encourage endless passive consumption of content on the Snapchat product. These features, in 8 tandem with each other and the other harmful features described throughout this section and 9 Complaint, induce addiction, compulsive use, and other severe mental and physical harm to the 10 child users of the Snapchat product, including Plaintiffs. 11 a. Snap designed Snapchat to drive compulsive use through a set of social metrics and other manipulation techniques that induce 12 compulsive use. 13 468. Snapchat includes a variety of social metrics—such as Snapscores, Snap Streaks, 14 and Snap Awards—that reward users when they engage with Snapchat and punish them when they 15 fail to engage with Snapchat. Internal research by Snap has found these psychological manipulation 16 techniques are highly effective at instilling anxiety about not using Snapchat frequently enough— 17 and competitor research has confirmed these features are addictive. In tandem with Intermittent and 18 Variable Rewards ("IVR"), like push notifications and design choices that make it difficult to stop 19 using the Snapchat product, these induce compulsive use of the product by children. 20 469. These manipulation techniques are so effective in part because Snapchat's 21 disappearing messages themselves create a compulsion to engage with the Snapchat product. 22 Because Snaps typically disappear within ten seconds of being viewed, users feel compelled to 23 reply immediately. Snap activates the psychological desire to reciprocate the social gesture of 24 sending a Snap.620 Snapchat also tells users each time they receive a Snap by pushing a notification 25 to the recipient's device. These notifications are designed to prompt users to open Snapchat 26 repetitively, increasing the overall time spent on the app. 27 620 28 Nir Eyal, The Secret Psychology of Snapchat, Nir & Far (Apr. 14, 2015), https://www.nirandfar.com/psychology-of-snapchat/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 152 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 159 of 278 1 (i) Snapscores 2 470. Snapscores were one of the earliest features of the Snapchat product. Almost as soon 3 as Snapchat launched, Snap gave users the ability to draw and color on Snaps and add a short text 4 caption before sending. An Android version of the app, video sharing, and user profiles with 5 "Snapscores" soon followed.621 6 471. Originally called "Hiscore," Snapscore keeps a running profile score based on a 7 user's Snapchat activity levels, such as the number of Snaps sent and received or Stories posted. 622 8 The sole purpose of Snapscore is to increase product use and drive revenue. 623 9 10 11 12 13 14 15 16 17 472. Although Snap does not disclose precisely how Snapscores work, sending and 18 receiving a Snap increases the score by one point. Interacting with other product features provides 19 additional points. A user's Snapscore is visible on their profile, serves as a signifier of the user's 20 "worth," and encourages users to further engage with Snapchat's features to increase their score. 21 Snapscores are important to users, especially young users, because they operate as a form of social 22 validation, similar to an Instagram "Like." Google has reported millions of searches for "How to 23 24 621 Snap Inc. Form S-1 Registration Statement (hereafter "Form S-1") at 91 (Feb. 2, 2017), https://www.sec.gov/Archives/edgar/data/1564408/000119312517029199/d270216ds1.htm; Katie 25 Notopoulos, The Snapchat Feature That Will You're your Life, BuzzFeed News (Dec. 5, 2012), � 622 Snapchat Support, What is a Snap Score?, ("Your Snapchat score is determined by a super- 26 secret, special equation…🤓") https://support.snapchat.com/en-US/a/my-score ("Your Snapchat score is determined by a super-secret, special equation… "). 27 623 Brad Barbz, *2020 NEW * How To Increase Snapscore By Up To 1000 Per Minute On IOS 28 And Android - Working 2020, YouTube (Dec. 4, 2019), https://www.youtube.com/watch?v=Mo_tajuofLA. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 153 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 160 of 278 1 improve Snap score." YouTube contains numerous videos with titles like "How to Increase 2 Snapchat Score Fast."624 3 473. Snapscores reward users who post videos that are viewed extensively. This 4 encourages many to use Snapchat in harmful and dangerous ways, to increase the virality of their 5 videos and increase their Snapscore. As more users engage with and forward that video to others, 6 its creator is awarded with an increased Snapscore. Snapchat's rewards incentivize this dangerous 7 behavior, resulting too often in physical harm or humiliation in the obsessive pursuit of social 8 significance. 9 (ii) Trophies, Charms, and Stickers 10 474. Snap has also designed Snapchat to include user rewards, including trophies and 11 other social recognition signals, similar to "Likes" on other apps. These features are highly 12 addictive and drive compulsive use. 13 475. "Trophies" are emojis awarded for achieving engagement milestones or performing 14 certain activities, such as increasing one's Snapscore, sending creative Snaps, or posting a live 15 story. A user's "Trophies" are displayed in a "trophy box" viewable by their friends. Snap designed 16 this feature to encourage users to share their videos and posts with the public, promote greater use 17 of Snapchat, and deepen young users' addiction to and compulsive use of the product. 18 476. In 2020, Snap phased out Trophies and replaced them with "Charms." Unlike 19 Trophies, where users were rewarded for unlocking individual accomplishments like sending 1,000 20 selfies, Charms reward users for achieving certain milestones in their relationship with other users. 21 Typically, the more users interact with one another, the more Charms they unlock in their 22 relationship. Charms are private and viewable only by users' mutual contacts. 23 477. For example, if two users are at the top of each other's friends list, meaning they 24 exchange frequent Snaps, they may unlock a "BFF (Best Friends Forever)" Charm. Conversely, 25 the "It's Been Forever" and "It's Been a Minute" Charms may be awarded to friends who are 26 infrequently in contact, to prompt their engagement with one another on Snapchat. Although there 27 624 FozTech, How to Increase Snapchat Score Fast! (100% Works in 2023), YouTube (Oct. 1, 28 2019), https://www.youtube.com/watch?v=m7s0hvQdTok (How to Increase Snapchat Score Fast has 4.2 million views as of January 10, 2023). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 154 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 161 of 278 1 are a number of different Charms awarded for various reasons, all of them encourage user 2 interaction, furthering engagement and buy-in to Snap's reward system. This in turn exacerbates 3 social-comparison harms and undermines self-esteem. 4 5 6 7 8 9 10 11 478. Snap incorporates other product features that, like Charms and Trophies, serve no 12 functional purpose, but make Snapchat more appealing and lead to excessive use by children and 13 teens. For example, Snap has developed images called "Stickers" for users to decorate the pictures 14 or videos they post. Snap also offers app-specific emojis and animations that users can apply to 15 their photos or videos. 16 479. Snap designed each of these features to function as rewards for increased 17 engagement, exploit underage users' desire for social validation, and ultimately compel them to use 18 Snapchat excessively. Because many of these rewards and scores are visible to others, these features 19 tap into adolescents' deep-seated need for acceptance. By exploiting this need, Snap increases time 20 spent engaging with its product and thereby its profits. 21 (iii) Snap Streak 22 480. The "Snap Streak" is unique to Snapchat and is an addictive feature "especially to 23 teenagers."625 A Snap Streak is designed to measure a user's Snapchat activity with another user. 24 625 See Cathy Becker, Experts warn parents how Snapchat can hook in teens with streaks, ABC 25 News (July 27, 2017), https://abcnews.go.com/Lifestyle/experts-warn-parents-snapchat-hook- teens-streaks/story?id=48778296; Avery Hartmans, These are the sneaky ways apps like 26 Instagram, Facebook, Tinder lure you in and get you 'addicted', Bus. Insider (Feb. 17 2018), https://www.businessinsider.com/how-app-developers-keep-us-addicted-to-our-smartphones- 27 2018-1#snapchat-uses-snapstreaks-to-keep-you-hooked-13; see generally Virginia Smart & Tyana Grundig, 'We're designing minds': Industry insider reveals secrets of addictive app trade, 28 CBC (Nov. 3, 2017), https://www.cbc.ca/news/science/marketplace-phones-1.4384876; Julian PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 155 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 162 of 278 1 Two users achieve a Snap Streak when they exchange at least one Snap in three consecutive 24- 2 hour periods. When the Streak is achieved, users receive a fire emoji next to their profile avatar. 3 Over time, users may be rewarded with additional emojis signifying their Streak. If users reach a 4 Streak of 100 days, for example, each receives a 100 emoji. 5 6 7 8 9 10 11 12 481. Snap Streak emojis are similar to Charms in that they reward users for interaction 13 and are viewable only by mutual friends. 14 482. It is a matter of common knowledge in the social media industry that the Snap Streak 15 product feature is designed to be addictive. Meta bluntly acknowledged as much in its internal 16 documents, stating: "Streaks are a very important way for teens to stay connected. They are usually 17 with your closest friends and they are addictive." 626 Nonetheless, Snap continues to provide this 18 feature to its adolescent users. 19 483. Worse still, to manufacture deeper addiction to its product, Snap sends notifications 20 to users with an hourglass emoji when Streaks are about to expire—to create extra urgency, nudge 21 users to keep their Streaks alive, and maintain a system where a user must "check constantly or risk 22 missing out."627 23 24 25 26 Morgans, The Secret Ways Social Media is Built for Addiction, Vice (May 17, 2017), https://www.vice.com/en/article/vv5jkb/the-secret-ways-social-media-is-built-for-addiction. 27 626 Haugen_00008303 at 8307. 627 28 Lizette Chapman, Inside the Mind of a Snapchat Streaker, Bloomberg (Jan. 30, 2017), https://www.bloomberg.com/news/features/2017-01-30/inside-the-mind-of-a-snapchat-streaker. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 156 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 163 of 278 1 2 3 4 5 6 7 8 484. This feature is particularly effective with teenage users since Streaks are "a vital part 9 of using the app and their social lives as a whole."628 Some children become so obsessed with 10 maintaining their Streaks that they give their friends access to their accounts when they may be 11 away from their phone for a day or more.629 Aware of how important maintaining a Snap Streak is 12 to its users, Snap has even launched a special form on its support website allowing users who lost 13 their streak to petition to get it back. 630 14 485. Snap Streaks contribute to feelings of social pressure and anxiety when users lose 15 or break a Streak. Researchers have found that losing a Streak can cause feelings of betrayal for 16 some users, especially girls, who reported "negative" feelings when losing a Streak with one of 17 their friends.631 18 486. In 2018, Snap conducted its own internal research on Snap Streaks, which found 19 that over a third of users reported it was "extremely" or "very important" to keep a Streak going, 20 21 628 22 Avery Hartmans, These are the sneaky ways apps like Instagram, Facebook, Tinder lure you in and get you 'addicted', Bus. Insider (Feb. 17, 2018), https://www.businessinsider.com/how- 23 app-developers-keep-us-addicted-to-our-smartphones-2018-1#snapchat-uses-snapstreaks-to-keep- you-hooked-13. 24 629 Caroline Knorr, How to resist technology addiction, CNN (Nov. 9, 2017), https://www.cnn.com/2017/11/09/health/science-of-tech-obsession-partner/index.html; Jon 25 Brooks, 7 Specific Tactics Social Media Companies Use to Keep You Hooked, KQED (June 9, 2017), https://www.kqed.org/futureofyou/397018/7-specific-ways-social-media-companies-have- 26 you-hooked. 630 Snapchat Support, Contact Form, https://support.snapchat.com/en-US/i-need- 27 help?start=5695496404336640. 631 28 Hristoya et al., "Why did we lose our snapchat streak?" Social media gamification and metacommunication. Computers in Human Behavior Reports, 5, 100172 (2022). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 157 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 164 of 278 1 and some users reported that the stress level to keep a Streak was "intolerable" or "large." Snap's 2 users reported that Streaks are equally important to Likes on Instagram. 632 3 487. As this research demonstrates, Streaks are important to users. However, these design 4 features do not enhance the communication function of the product. Instead, they exploit users' 5 susceptibility to social pressure and to the compulsive accumulation of other rewards, including 6 Snap Score points and Charms. 7 (iv) Push Notifications 8 488. In addition to Snapchat's in-app reward features, Snap also sends push notifications 9 and emails to encourage addictive engagement and increase use. Notifications are triggered based 10 on information Snap collects from, and about, its users. Snap "pushes" these communications to 11 users excessively and at disruptive times of day. Snap has even designed the format of these 12 notifications to pull users back onto its app by preying on their fear of missing out—never mind 13 the consequences to their health and well-being. 14 (v) Impediments to Discontinuing Use 15 489. Snap has intentionally and defectively designed its products so child users face 16 significant navigational obstacles and hurdles when trying to delete or deactivate their Snapchat 17 accounts, despite the ease with which a user can create one. For example, when a user elects to 18 delete their account, they cannot do so on demand. The data and the account are preserved for 30 19 days. In addition, after initiating the deletion process, the user is presented with a black screen 20 depicting a crying emoji and a message that reads, "Your account will be deactivated, which means 21 friends won't be able to contact you on Snapchat. You'll also lose any Chats you've saved and 22 Snaps and Chats you haven't opened."633 23 490. This cumbersome process prioritizes user retention and continued use over the well- 24 being of Snapchat's users. 25 26 27 632 SNAP0000008. 633 28 See Snapchat Support, How do I delete my Snapchat account?, https://support.snapchat.com/en-US/a/delete-my-account1 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 158 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 165 of 278 1 b. Snap's defective features are designed to promote compulsive and excessive use. 2 (i) "Stories" and the "Discover" Interface 3 4 491. In October 2013, Snap added "Stories," a feature that generates a compilation of its 5 users' designated photos and videos that expire in 24 hours and can be viewed an unlimited number 6 of times by friends or anyone on Snapchat if the user sets the visibility setting to Everyone. 634 7 Within eight months of launching the Stories feature, users were viewing more Stories per day than 8 Snaps.635 9 492. Snap's Stories feature includes a running view count and list of viewers for each 10 Story, both of which provide users with dopamine-triggering feedback that encourages users to 11 make their Stories visible to everyone in order to increase the view count. The view count, view 12 list, and ephemeral nature of Stories also reinforces the principle of reciprocity and compels users 13 to monitor Stories, so they do not miss out. 14 493. In 2016, Snap updated the Stories feature to include recommendations based on an 15 algorithm that considers "proximity, time, interestingness, or other such metrics." 636 That same 16 year, Snap introduced ads between Stories and updated Stories to include "Auto-Advance," a 17 feature that starts a new Story automatically after the preceding one ends. 637 This creates an endless 18 cycle of consumption that Snap knows, or should know, is detrimental to users' mental health. 638 19 Nevertheless, Snap designed and implemented this feature because it is proven to induce a flow 20 634 21 Darrell Etherington, Snapchat Gets Its Own Timeline With Snapchat Stories, 24-Hour Photo & Video Tales, TechCrunch (Oct. 3, 2013), https://techcrunch.com/2013/10/03/snapchat-gets-its- 22 own-timeline-with-snapchat-stories-24-hour-photo-video-tales/. 635 Ellis Hamburger, Surprise: Snapchat's most popular feature isn't snaps anymore, The Verge 23 (Jun. 20, 2014), https://www.theverge.com/2014/6/20/5827666/snapchat-stories-bigger-than- snaps-electric-daisy-carnival 24 636 Snapchat, Inc., Content Collection Navigation and Autoforwarding, US 20170289234, USPTO (Mar. 29, 2016), https://patents.justia.com/patent/20170289234. 25 637 James Vincent, Snapchat will start showing ads between your friends' stories, The Verge (Jun. 14, 2016), https://www.theverge.com/2016/6/14/11930386/snapchat-ads-api-stories; Snapchat, 26 Inc., Content Collection Navigation and Autoforwarding, US 20170289234, USPTO (Mar. 29, 2016), https://patents.justia.com/patent/20170289234. 27 638 See, e.g., Gino Gugushvili et al., Facebook use intensity and depressive symptoms: a 28 moderated mediation mode of problematic Facebook use, age, neuroticism, and extraversion at 3, BMC Psych. 10, 279 (2022), https://doi.org/10.1186/s40359-022-00990-7. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 159 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 166 of 278 1 state that increases product use, regardless of whether the use is healthy or enjoyable. 2 Unsurprisingly, one study of over 2,000 UK residents found 68% of respondents who used 3 Snapchat reported that "the platform prevented them from sleeping." 639 4 494. Since then, Snap has built upon its Stories interface with "Discover," a feature that 5 showcases a continuous feed of advertisements to Snapchat's captive audience. Using Discover, 6 users may subscribe to an advertiser's "channel" and watch its Stories; as well as see what their 7 friends are watching. 8 495. Both Stories and Discover encourage user engagement with Snapchat and increase 9 the amount of time users spend using the product by making the product more addictive at the 10 expense of users' mental health and well-being. 11 (ii) "Spotlight' 12 496. In November 2020, Snap launched "Spotlight," a feature that pushes to users "an 13 endless feed" that Snap curates from its 300 million daily Snapchat users. 640 Spotlight functions 14 and appears nearly identical to TikTok, with similar addictive qualities and harms. Snapchat's 15 Spotlight feature allows users to make videos that anyone can view, and Snap pays users whose 16 Spotlight videos go viral, thus serving as yet another reward system that encourages user 17 engagement. After Snap introduced Spotlight, user time spent on the product increased by over 18 200%.641 19 497. In February 2022, Snap CEO Evan Spiegel told investors that users are spending 20 more time on Spotlight than almost any other aspect of Snapchat. A year prior, Snap announced 21 "Spotlight Challenges," which provided users with cash prizes for creating Spotlight videos with 22 specific lenses, sounds, or topics, further integrating the user into the Snap ecosystem. Snap claims 23 24 25 639 Frazer Deans, Curb Your Snapchat Addiction, https://www.wholesome.design/advent-2018/2- curb-your-snapchat-addiction/. 26 640 Salvador Rodriguez, Snap is launching a competitor to TikTok and Instagram Reels, CNBC (Nov. 23, 2020), https://www.cnbc.com/2020/11/23/snap-launching-a-competitor-to-tiktok-and- 27 instagram-reels.html. 641 28 See Snap Q4 Earnings Beat Estimates, User Growth Aids Top Line, Zacks Equity Research (Feb. 5, 2021), https://finance.yahoo.com/news/snap-q4-earnings-beat-estimates-153003950.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 160 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 167 of 278 1 it paid out more than $250 million in cash prizes to Spotlight Challenge participants in 2021 2 alone.642 3 4. Snap designed Snapchat with features that harm children directly or expose children to harm. 4 5 498. Snapchat further contains a number of features which foreseeably cause children 6 harm above and beyond harms inherent in addiction and compulsive use. 7 a. Disappearing "Snaps" and "My Eyes Only" encourage destructive behavior among Snap's teen users. 8 9 499. As discussed above, Snapchat's "Snap" feature allows users to send and receive 10 ephemeral, or "disappearing," audiovisual messages. Prior to sending a Snap, a user can designate 11 the period of time—typically no more than a few seconds—that the recipient will be allowed to 12 view the Snap. According to Snapchat, once the allotted time expires, the Snap disappears forever. 13 500. Snapchat's limited display time reduces teenagers' communication apprehension 14 and encourages users to send photos depicting deviant behavior. 643 Sexting is a prime example, but 15 cyberbullying, underage alcohol consumption, and illicit use of narcotics are also commonly the 16 subject of Snaps. A 2016 survey of pre-teens and teens ages 12-17 found that "dick pics" were 17 among some of the unwanted content that users—predominantly females—received while using 18 the app.644 19 501. Disappearing Snaps do not operate as advertised. Although designed to disappear 20 after an allotted time, recipients possess the ability to save or record them at will. This is particularly 21 harmful to adolescents, who rely on Snap's representations when taking and sending photos, and 22 who only learn after the fact that recipients have the means to save photos or videos. In some cases, 23 this can lead to sexual exploitation. 24 25 642 Mia Sato, Snapchat will put ads within stories and share the money with creators (Feb. 14, 2022), https://www.theverge.com/2022/2/14/22927656/snapchat-snap-stars-stories-ads. 26 643 See Vaterlaus et al., "Snapchat is more personal": An exploratory study on Snapchat behaviors and young adult interpersonal relationships, Computers in Human Behavior, 62, 594- 27 601 (2016). 644 28 Kofoed et al., (2106) A snap of intimacy: Photo-sharing practices among young people on social media, First Monday 21(11), https://doi.org/10.5210/fm.v21i11.6905. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 161 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 168 of 278 1 502. Snap could, but does not, warn users, including children and teenagers, that Snaps 2 may not necessarily disappear. 3 503. In addition, and especially for pre-teen users, Snaps are defective because Snap's 4 parental controls are ill-equipped to mitigate the risks posed by this feature. As set forth below, 5 even with parental controls activated, parents are unable to view a Snap's content and therefore 6 cannot adequately protect their children and/or deter their children from engaging in dangerous 7 behavior in conjunction with sending Snaps. 8 504. "My Eyes Only" is yet another defective feature of Snapchat. This feature enables 9 and encourages users to hide harmful content from their parents in a special tab that requires a 10 passcode. Content cannot be recovered from "My Eyes Only"—allegedly even by Snap itself. Snap 11 designed "My Eyes Only" knowing it would likely be used to store potentially illegal and injurious 12 photos and images like sexts and CSAM.645 This dangerous product feature unreasonably increases 13 the risk that Snapchat's adolescent users, many under age 13, will be targeted and sexually 14 exploited and/or trafficked by child predators. 15 505. The content in "My Eyes Only" self-destructs if a user attempts to access the hidden 16 folder with the wrong code. "My Eyes Only" has no practical purpose or use other than to hide 17 potentially dangerous content from parents and/or legal owners of the devices used to access 18 Snapchat. Moreover, while this information and evidence should be in Snap's possession and 19 control, it has designed this feature in a way that causes the permanent loss of relevant, material, 20 and incriminating evidence. 21 b. Snapchat's "Snap Map" feature endangers children. 22 506. Snapchat also contains a feature called "Snap Map" that allows users to share their 23 location with their followers (and the public) on an activity-level-based, color-coded heatmap. At 24 all relevant times, this feature has been available to all users, including minors. Although users can 25 disable "Snap Map," this is not a default setting. 26 645 Salvador Rodriguez, Snapchat Finally Acknowledges the Existence of Sexting With 'Memories' 27 The latest app update includes a tool called "My Eyes Only" that lets you privately store sensitive photos and videos, Inc., (Jul. 6, 2016), https://www.inc.com/salvador-rodriguez/snapchat- 28 memories-sexting.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 162 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 169 of 278 1 507. Snap Map is an unreasonably dangerous feature for underage users because it 2 provides strangers with their locations, exposing children and adolescents to potential harm. 3 Researchers have also found that Snap Map causes feelings of sadness and anxiety for some users, 4 as they jealously view their friends' locations. 646 For young people especially, such social 5 comparison often leads to distress and depression. 6 508. Snap Map also functions as a social metric. A report by 5Rights, a United Kingdom- 7 based children's online safety advocacy group highlighted the experience of John, a 14 year old 8 boy, who explained that "[h]aving more connections on Snapchat makes his Snap Map look more 9 crowded, which he can then show off to people in real life and therefore appear more 'popular.'" 647 10 c. Snapchat's "Quick Add" feature endangers children. 11 509. Through a feature known as "Quick Add," Snap recommends new, sometimes 12 random friends, similar to Facebook's "People You Might Know" feature. Suggestions are 13 formulated using an algorithm that considers users' friends, interests, and location. Quick Add 14 encourages users to expand their friend base to increase their Snapscore by interacting with an ever- 15 expanding group of friends, which--in addition to expanding their time online--can result in 16 exposure to dangerous strangers. Of particular concern, until 2022, Quick Add's suggestions 17 included profiles for users Snap knew to be between the ages of 13-17, meaning that Quick Add 18 could, and in fact did, recommend that a minor and adult user connect. 19 510. Criminal users interested in selling drugs to minors have utilized the Quick Add 20 feature to find random friends interested in making a purchase. 21 511. Despite these dangers Snap designed Quick Add because it increases the odds that 22 users will add friends, send more Snaps, and spend more time using Snapchat. 23 512. In 2022, Snap revised the Quick Add feature to limit the friend suggestions 24 promoted to minor users. For those aged 13 to 17, Quick Add would only suggest friends who 25 shared a certain number of common friends with the minor user. Snap did not disclose how many 26 646 See Dunn et al., "Oh, Snap!": A Mixed-Methods Approach to Analyzing the Dark Side of 27 Snapchat, The Journal of Social Media in Society, 9(2), 69-104 (2020). 647 28 5Rights Foundation, Pathways: How digital design puts children at risk (July 2021), https://5rightsfoundation.com/uploads/Pathways-how-digital-design-puts-children-at-risk.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 163 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 170 of 278 1 common friends must be shared by each user to satisfy this safety feature. Further, this modification 2 to the Quick Add feature still does not prohibit the connection of minors with adults. 3 d. Snapchat's Lenses and Filters features promote negative appearance comparison. 4 513. Snap also incorporates numerous custom-designed lenses and filters, which allow 5 users to edit and overlay augmented-reality special effects and sounds on their Snaps. Many of 6 Snapchat's lenses and filters change users' appearance and face, creating unrealistic, idealized 7 versions that cause profound body image issues in teenagers, especially girls. 8 514. Examples of these features include the Smoothing Filter, which blurs facial 9 imperfections and evens out skin tone; Bold Makeup, which adds makeup over the user's face, 10 blurs imperfections, and evens out skin tone; Sunkissed and Cute Freckles, which adds freckles 11 over the nose and cheeks, blurs imperfections, evens out skin tone, and adjusts skin color; Face and 12 Body Mellow Glow, which smooths the face and body and adjusts skin color; and Fluffy Eyelashes, 13 which alters the shape of the user's face by lifting their eyes and adding more pronounced cheek 14 bones. The common theme among all of these filters is that they remove the subjects' perceived 15 blemishes to create the perfect "selfie." 16 17 18 19 20 21 22 23 24 515. A 2017 study found that these features made Snapchat one of the worst social media 25 products for the mental health of children and adolescents, behind only Instagram. 648 In recent 26 years, plastic surgeons have reported an increase in requests for alterations that correspond to 27 648 Kara Fox, Instagram worst social media app for young people's mental health, CNN (May 19, 28 2017), https://www.cnn.com/2017/05/19/health/instagram-worst-social-network-app-young- people-mental-health/index.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 164 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 171 of 278 1 Snapchat's filters. This has led researchers to coin the term "Snapchat Dysmorphia," in which the 2 effect of Snapchat's filters triggers body dysmorphic disorder. 649 The rationale underlying this 3 disorder is that beauty filters on Snapchat create a "sense of unattainable perfection" that leads to 4 self-alienation and damages a person's self-esteem.650 One social psychologist summarized the 5 effect as "the pressure to present a certain filtered image on social media," which can certainly play 6 into [depression and anxiety] for younger people who are just developing their identities." 651 7 8 9 10 11 12 13 516. Snap also created and promoted "smart filters" that allowed users to stamp 14 date/time, temperature, battery life, altitude, and speed on their Snaps. 652 These filters utilize sensor 15 data on users' devices to provide the desired filter stamp. 16 517. A particularly dangerous smart filter is the speed filter, which from 2013 to 2021 17 allowed users to record their real-life speed and overlay that speed onto Snaps. Snap knew, or 18 should have known, that the speed filter served no purpose other than to motivate, incentivize, 19 and/or encourage users to drive at dangerous speeds in violation of traffic and safety laws. Indeed, 20 649 21 Chen et al., Association Between Social Media and Photograph Editing Use, Self-esteem, and Cosmetic Surgery Acceptance, JAMA Facial Plastic Surgery, 2019; See also Nathan Smith & 22 Allie Yang, What happens when lines blur between real and virtual beauty through filters, ABC News (May 1, 2021), https://abcnews.go.com/Technology/lines-blur-real-virtual-beauty- 23 filters/story?id=77427989. 650 Chen et al., Association Between Social Media and Photograph Editing Use, Self-esteem, and 24 Cosmetic Surgery Acceptance, JAMA Facial Plastic Surgery, 2019; See also Nathan Smith & Allie Yang, What happens when lines blur between real and virtual beauty through filters, ABC 25 News (May 1, 2021), https://abcnews.go.com/Technology/lines-blur-real-virtual-beauty- filters/story?id=77427989. 26 651 Nathan Smith & Allie Yang, What happens when lines blur between real and virtual beauty through filters, ABC News (May 1, 2021), https://abcnews.go.com/Technology/lines-blur-real- 27 virtual-beauty-filters/story?id=77427989. 652 28 Karissa Bell, Snapchat adds an altitude filter to show how high you are, (Aug.19, 2016), https://mashable.com/article/snapchat-altitude-filter-how-to. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 165 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 172 of 278 1 soon after launching its speed filter, the feature became a viral game for users—particularly teenage 2 users—to capture photos and videos of themselves driving at 100 miles-per-hour or more. 3 Tragically, the quest to capture a 100 mile-per-hour Snap caused a number of fatal vehicle accidents 4 involving teens and young adults.653 5 518. Snap knew, or should have known, its speed filter created an unreasonable risk of 6 harm to its users and the public. Despite this knowledge, however, as well as pleas from the public 7 to disable the filter, Snap refused to remove the filter from its application until 2021. 654 8 519. By including features like lenses, cartoonish filters, and stamps to attract ever- 9 increasing numbers of children to use and engage with its product, Snap has knowingly created a 10 product that leads to excessive use by children and teens and causes them to suffer harm. 11 5. Snap has implemented ineffective and misleading parental controls, further endangering children. 12 13 520. Snap has also designed and set up Snapchat with inadequate parental controls. 14 521. From Snapchat's launch in 2011 until August 2022, Snapchat had no parental 15 controls even though its core user base was under the age of 18 and a significant number of those 16 users were under the age of 13. 17 522. In August 2022, Snap introduced the "Family Center." The features and processes 18 offered through the Family Center are woefully inadequate to protect teen and pre-teen users. The 19 Family Center allows a parent or guardian to install Snapchat on their phone and then link to the 20 child's account. The parent or guardian can then see who the child user is communicating with. 21 However, the content of these communications remains hidden and still disappears after the allotted 22 time. In addition, the Family Center does not allow a parent or guardian to block minors from 23 653 Did Snapchat play role in deaths of 3 young women?, ABC6 Action News (Feb. 16, 2016), 24 https://6abc.com/action-news-investigation-snapchat-fatal-car-crash-philadelphia/1196846/; Manpreet Darroch, Snapchat and driving . . . you could be sending your last snap (Apr.25, 2016), 25 http://www.youthforroadsafety.org/news-blog/news-blog-item/t/snapchat-and-driving-hellip-you- could-be-sending-your-last-snap; The Most Dangerous App on Your Phone, 26 DistractedDriverAccidents.com, https://distracteddriveraccidents.com/the-most-dangerous-app- on-your-phone/. 27 654 Bobby Allyn, Snapchat Ends 'Speed Filter' That Critics Say Encouraged Reckless Driving, 28 NPR (June 17, 2021), https://www.npr.org/2021/06/17/1007385955/snapchat-ends-speed-filter- that-critics-say-encouraged-reckless-driving. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 166 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 173 of 278 1 sending private messages, control their child's use or engagement with many of Snapchat's product 2 features, control their child's use of Snapchat's geolocation feature, or control who their child may 3 add to their friend list. Finally, the Family Center fails to help a parent monitor their child's account 4 when the child has secretly created a Snapchat account without the parents' knowledge in the first 5 place. 6 6. Snap facilitates the spread of CSAM and child exploitation. 7 523. Despite being marketed to and designed for children, Snapchat includes a number 8 of features that promote and dramatically exacerbate sexual exploitation, the spread of CSAM, 9 sextortion, and other socially maladaptive behavior that harms children. Snap knows or should have 10 known that its product features are unsafe for children and that it fails to implement reasonable, 11 child-protective safeguards. For example, by failing to age-restrict its Discover feature, Snapchat's 12 algorithm has recommended inappropriate sexual content to adolescent users. By promoting the 13 connection between minors and adults, it is facilitating child exploitation and predation. By failing 14 to implement parental controls that give parents true control over their children's activity, Snap 15 allows harmful interactions with predators to continue unnoticed. 16 524. Like the other Defendants, as a direct consequence of the child exploitation that 17 occurs on its platform, Snapchat is tainted by illegal material that promotes and facilitates the 18 continued sexual exploitation of minors. Snap receives value in the form of increased user activity 19 for the dissemination of CSAM on its product. 20 525. Furthermore, Snapchat's disappearing-content design, while appealing to minors, 21 makes it more difficult for parents to monitor their children's social-media activity. This feature 22 also contributes to a sense of impunity for many users, encouraging and fomenting exploitation and 23 predatory behavior, which has been observed in multiple empirical studies. 655 According to these 24 studies, Snapchat users believe their conduct is hidden and accordingly feel empowered to engage 25 in criminal behavior through the product without fear of getting caught. 26 27 655 28 Snapchat by the Numbers: Stats, Demographics & Fun Facts, Omnicore (Mar. 2, 2022), https://www.omnicoreagency.com/snapchat-statistics/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 167 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 174 of 278 1 526. These feelings are promoted by design. Snap intends for the product's disappearing 2 messaging to entice users to share highly personal photos and information that many users would 3 otherwise feel uncomfortable sharing on "higher-stake" apps. 656 In short, this design choice 4 encourages and allows minors to share harmful, illegal, and sexually explicit images while 5 providing predators with a vehicle to recruit victims. Studies have also found that the "close ties" 6 generated between teenagers on Snapchat foster the conditions for grooming and other predatory 7 behavior. 8 527. As a result, Snapchat is one of the go-to products for sexual predators. 657 9 528. In 2014, Snap introduced "Snapcash," a peer-to-peer mobile payment service. 10 Snapcash provided a way for users to pay for private content with little to no oversight. 658 Snapcash 11 enabled CSAM and other sexual exploitation, as users were paid with Snapcash to send, receive, 12 create, publish, save, accept, or otherwise participate in CSAM. It also enabled predators to extort 13 cash from adolescent users by threatening to disseminate CSAM to other users. 14 529. Snapcash was abruptly removed from Snapchat in 2018 as users were sending 15 sexually explicit photos and using Snapcash for payment. 659 16 530. Snapchat also allows users to voice or video call one another in the app. 660 This 17 feature is dangerous when paired with the many others that permit easy access to minors by 18 predators, such as Quick Add and Snap Map. It allows predators to call and video chat with minors 19 in private, with virtually no evidence of what was exchanged. Predators use this function to identify 20 children willing to add and speak with a stranger, and then prey on the child's vulnerabilities. 21 656 See Evelyn Lopez et al., The Gratifications of Ephemeral Marketing Content, the Use of 22 Snapchat by the Millenial Generation and Their Impact on Purchase Motivation, Global Bus. Rev. (2021), https://journals.sagepub.com/doi/pdf/10.1177/09721509211005676. 657 23 See, e.g., Rebecca Woods, What Are The Dangers Of Snapchat To Avoid?, PhoneSpector (June 16, 2021), https://phonespector.com/blog/what-are-the-dangers-of-snapchat-to-avoid/. 24 658 Kurt Wagner, Snapchat to Let You Send Money to Friends, Thanks to Square, Vox, https://www.vox.com/2014/11/17/11632930/snapchat-to-let-you-send-money-to-friends-thanks- 25 to-square. 659 Christian Hargrave, Snapcash Goes Away After Excessive Feature Misuse. App Developer 26 Magazine (July 25, 2018), https://appdevelopermagazine.com/snapcash-goes-away-after- excessive-feature-misuse/. 27 660 Snapchat Support, How to Start a Video Chat on Snapchat, https://support.snapchat.com/en- 28 GB/a/video- chat#:~:text=You%20can%20Video%20Chat%20with,into%20a%20full%2Dscreen%20Chat. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 168 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 175 of 278 1 531. Collectively, these product features promulgate communication and conduct with a 2 false sense of intimacy between users and encourage predators to use Snapchat to target children 3 for grooming, sexual exploitation, sextortion, and CSAM. 4 532. In November 2019, a bipartisan group of Senators sent a letter to leading tech 5 companies, including Snapchat. The letter sought answers about the online sexual grooming of 6 children and CSAM detection technologies. 661 The following year, ParentsTogether, a national 7 parent group, delivered a petition from 100,000 parents to Snap demanding that the company do 8 more to "protect children from sexual abuse and exploitation" on Snapchat. 662 The petition listed 9 numerous examples of widespread online sexual grooming of children, including: a high school 10 coach in New Mexico who used Snapchat to extort sexual videos from several girls as young as 11 fourteen; a Cleveland man who posed as a therapist and blackmailed a thirteen-year-old girl into 12 sending him sexual videos and photos; and a Virginia man who was arrested for running a 13 sextortion ring on Snapchat, coercing children into sending sexually explicit material. 663 14 533. In response, Snap announced that by Fall of 2020, it would deploy technology in 15 addition to Microsoft's PhotoDNA to help stop the spread of CSAM through its product. 16 534. By failing to utilize these technologies until late 2020, Snap harmed adolescent users 17 as its product contributed to child exploitation, sextortion, and the spread of CSAM. 18 535. In addition, while Snapchat allows users to report harmful images or videos, they 19 cannot specifically report CSAM that is sent to a user via direct messaging, including from another 20 user's camera roll. 21 536. Snapchat's disappearing messages cannot be reported at all. 22 537. While Snap states that it is using "technology to identify known illegal images and 23 videos of CSAM and report them to NCMEC," it does not address how Snapchat's design 24 661 Letter to Sundar Pichai and 36 other Tech Companies by Senate Committee (Nov. 18, 2019), 25 https://www.blumenthal.senate.gov/imo/media/doc/11.18.19%20-%20Google%20- %20CSAM.pdf. 26 662 Snapchat: Prevent Pedophiles from Sharing Abuse Videos, https://parents- together.org/snapchat-petition. 27 663 Snapchat: Prevent Pedophiles from Sharing Abuse Videos, https://parents- together.org/snapchat-petition. 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 169 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 176 of 278 1 contributes to the ongoing proliferation of CSAM materials and the sexual exploitation of its 2 adolescent users. 3 538. Utilizing the data and information it collects about Snapchat's users, Snap could 4 detect, report, and take actions to prevent instances of sexual grooming, sextortion, and CSAM 5 distribution. 6 539. Despite receiving numerous reports regarding how its product's features contribute 7 to child exploitation, Snap has elected to keep many of these features in place. 664 It has done so 8 because removing them would significantly diminish Snapchat's popularity and negatively impact 9 profits. 10 540. Notwithstanding these glaring flaws, Snap advertises and promotes its product as 11 safe and fun. Snap's Vice President of Global Public Policy, Jennifer Stout, stated in written 12 testimony to a Senate Subcommittee that Snap takes "into account the unique sensitivities and 13 considerations of minors when we design products"665 when, in fact, Snap intentionally designed 14 its product to promote compulsive and excessive use and help underage users conceal information 15 from their parents. Stout claimed that Snap makes it harder for strangers to find minors when, in 16 fact, Snapchat's "Quick Add" feature is responsible for introducing minors to complete strangers, 17 and its "Snap Map" feature has enabled threats, exploitation, and location of minors by complete 18 strangers. Likewise, Snap's Head of Global Platform Safety, Jacqueline Beauchere, represented to 19 the public that "Snapchat is designed for communications between and among real friends; it 20 doesn't facilitate connections with unfamiliar people like some social media platforms." 666 But 21 again, this is not true and/or historically was not the case. 22 23 24 664 See, e.g., Zak Doffman, Snapchat Has Become A 'Haven For Child Abuse' With its "Self- Destructing Messages', Forbes (May 26, 2019), 25 https://www.forbes.com/sites/zakdoffman/2019/05/26/snapchats-self-destructing-messages-have- created-a-haven-for-child-abuse/?sh=411b8e1d399a. 26 665 Snap's Senate Congressional Testimony - Our Approach to Safety, Privacy and Wellbeing, https://values.snap.com/news/senate-congressional-testimony-our-approach-to-safety-privacy- 27 and-wellbeing. 666 28 Snap's Meet Our Head of Global Platform Safety, https://values.snap.com/news/meet-our- head-of-global-platform-safety. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 170 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 177 of 278 1 541. In addition, Snap knows or should have known, that its products facilitate and 2 encourage the production, possession, distribution, receipt, transportation, and dissemination of 3 millions of materials that exploit children and violate child pornography laws. Snap further knows, 4 or should have known, that its product facilitates the production, possession, distribution, receipt, 5 transportation, and dissemination of materials that depict obscene visual representations of the 6 sexual abuse of children. 7 542. Upon information and belief, Snap has developed, or is developing, artificial 8 intelligence technology that detects adult users of Snapchat who send sexually explicit content to 9 children and receive sexually explicit images from children. This technology furnishes Snap with 10 actual knowledge that a significant number of minor users of Snapchat are solicited to send, and do 11 send, sexually explicit photos and videos of themselves to adult users in violation of 18 U.S.C. 12 § 1591(a)(1)-(2).667 13 7. Snap failed to adequately warn Plaintiffs about the harms its product causes or provide instructions regarding safe use. 14 15 543. Since Snap's inception, it has failed to warn adolescent users about its products' 16 physical and mental health risks. These risks include, but are not limited to, addiction, compulsive 17 and excessive use, sexual exploitation by adult users, dissociative behavior, social isolation, and an 18 array of mental health disorders like body dysmorphia, anxiety, depression, and insomnia. 19 544. Snap targets adolescent users via advertising and marketing materials distributed via 20 digital and traditional media, including expensive advertisements placed during high-profile 21 sporting events. Snap fails to warn the targets of these ads—often minors—about the physical and 22 mental risks associated with using Snapchat. 23 545. Snap further fails to warn adolescent users during the product registration process. 24 At account setup, Snap's product contains no warning labels, banners, or conspicuous messaging 25 to adequately inform adolescent users of the known risks and potential physical and mental harms 26 associated with usage of its product. Instead, Snap allows adolescent users to easily create an 27 account (or multiple accounts) and fully access the product. 28 667 See SNAP0000001-SNAP0000002. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 171 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 178 of 278 1 546. Snap's lack of adequate warnings continues after an adolescent has the Snapchat 2 product. Snap does not adequately inform adolescent users that their data will be tracked, used to 3 help build a unique algorithmic profile, and potentially sold to Snap's advertising clients, who will 4 in turn use the data to target and profile the user. 5 547. Alarmingly, Snap also does not warn adolescent users before facilitating adult 6 connections and interactions that adult predators use its product. It also fails to instruct adolescent 7 users on ways to avoid unknown adults on Snap. 8 548. Snap also fails to warn adolescent users who exhibit problematic signs of addiction 9 or are habitually and compulsively accessing the app. Instead, Snap utilizes push notifications to 10 encourage engagement with Snapchat. 11 549. In addition, despite proactively providing adolescent users with countless filtering 12 and editing tools, Snap does not warn its adolescent users regarding the mental health harms 13 associated with those heavily filtered images. 14 550. Snap also does not warn users that: 15 adult predators target children and teenagers for sexual exploitation, 16 sextortion, and CSAM on Snap's product; 17 usage of its product can increase the risk of children being targeted and 18 sexually exploited by adult predators; 19 usage of its product can increase risky and uninhibited behavior in children, 20 making them easier targets for sexual exploitation, sextortion, and CSAM 21 by adult predators; and, 22 end-to-end encryption and/or ephemeral nature of Snap's direct messaging 23 product prevents the reporting of CSAM. 24 551. Snap's failure to properly warn and instruct adolescent users has proximately caused 25 significant harm to Plaintiffs' mental and physical well-being, and other injuries and harms as set 26 forth herein. 27 552. Snap also fails to warn parents about all of the foregoing dangers and harms. 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 172 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 179 of 278 1 553. Snap's failure to adequately warn and instruct as set forth herein proximately caused 2 significant harm to Plaintiffs' mental and physical well-being, and other injuries and harms as set 3 forth herein. 4 D. FACTUAL ALLEGATIONS AS TO BYTEDANCE 5 554. Since its launch, TikTok has grown exponentially. In late 2021, its owner and creator 6 ByteDance publicly stated that TikTok had 1 billion active global users, up from 55 million in early 7 2018 and 700 million in mid-2020.668 One internal marketing strategy document reported that as of 8 July 2020, TikTok had 95% market penetration in Mobile App Users/ smartphone population of 9 users under 17.669 10 555. A large swath of TikTok's user base is comprised of American children. In July 11 2020, TikTok reported that more than one-third of its 49 million daily users in the United States 12 were 14 or younger.670 More recently, a 2022 Pew Research Center survey reported that 67% of 13 American teenagers (age 13-17) use TikTok, with most American teenagers (58%) using the 14 product daily. Among teenage TikTok users, a quarter say they use the site or app almost 15 constantly.671 In another recent report, more than 13% of young users declared they "wouldn't want 16 to live without" TikTok.672 17 556. TikTok's capture of the American youth market is no accident, but instead the result 18 of a carefully executed campaign. Early on, Alex Zhu, one of TikTok's creators, recognized that 19 "[t]eenagers in the U.S. [were] a golden audience" for this emerging social media product. 673 To 20 668 Jessica Bursztynsky, TikTok says 1 billion people use the app each month, CNBC (Sept. 27, 21 2021), https://www.cnbc.com/2021/09/27/tiktok-reaches-1-billion-monthly-users.html. 669 22 TIKTOK3047MDL-001-00000622 at *27. 670 Raymond Zhong & Sheera Frenkel, A Third of TikTok's U.S. Users May Be 14 or Under, 23 Raising Safety Questions, N.Y. Times (Aug. 14, 2020), https://www.nytimes.com/2020/08/14/technology/tiktok-underage-users-ftc.html. 24 671 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 25 2022/. 672 Victoria Rideout et al., Common Sense Census: Media use by tweens and teens, 2021 at 31, 26 Common Sense Media (2022), www.commonsensemedia.org/sites/default/files/research/report/8- 18-census-integrated-report-final-web_0.pdf. 27 673 Paul Mozur, Chinese Tech Firms Forced to Choose Market: Home or Everywhere Else, N.Y. 28 Times (Aug. 9, 2016), https://www.nytimes.com/2016/08/10/technology/china-homegrown- internet-companies-rest-of-the-world.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 173 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 180 of 278 1 cash in on this gold, ByteDance implemented a series of product features designed to attract and 2 addict young users. As Zhu explained in 2019, "[e]ven if you have tens of millions of users, you 3 have to keep them always engaged." 674 This engagement has come at the cost of young users' 4 health. 5 1. Background and overview of TikTok. 6 557. In 2012, Beijing-based technologist Zhang Yiming paired up with an American 7 venture capitalist, Matt Huang, to launch ByteDance and its first product Jinri Toutiao ("Today's 8 Headlines"), which utilized A.I. to gather and present world news to users on a single feed. 9 558. Following the success of its first product, ByteDance created Douyin in 2016, a 10 music-based app loosely modeled on the popular app Musical.ly. Musical.ly was a critical hit in 11 the U.S. as American teens gravitated to the platform, which allowed users, including minor users, 12 to create 15-second videos of themselves lip-syncing, dancing, or goofing around to popular songs 13 and movie scenes, and then post them to a scrollable feed for other users to see. 14 559. In 2017, ByteDance launched TikTok, a version of Douyin for the non-Chinese 15 market, and acquired Musical.ly, which by then boasted a user base of almost 60 million monthly 16 active users, for $1 billion. Nine months later, ByteDance merged its newly acquired app into its 17 existing product, and a global version of TikTok was born. 18 560. Douyin is a version of TikTok that is exclusively for Chinese users. ByteDance's 19 design of Douyin is profoundly different than TikTok. Douyin serves its Chinese users educational 20 and patriotic content and limits their use to just 40 minutes per day.675 TikTok, in sharp contrast, 21 has no usage limits and is designed to encourage addictive and compulsive use. Far from promoting 22 educational content, TikTok's algorithm instead actively sends its young American users down a 23 harmful rabbit hole of artificially filtered "ideal" body images and dangerous viral challenges. 24 25 674 Biz Carson, How A Failed Education Startup Turned into Musical.ly, The Most Popular App 26 You've Probably Never Heard Of, Bus. Insider (May 28, 2016), https://www.businessinsider.com/what-is-musically-2016-5 (emphasis added). 27 675 Sapna Maheshwari, Young TikTok Users Quickly Encounter Problematic Posts, Researchers 28 Say, N.Y. Times (Dec. 14, 2022), https://www.nytimes.com/2022/12/14/business/tiktok-safety- teens-eating-disorders-self-harm.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 174 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 181 of 278 1 2. ByteDance intentionally encourages youth to use its product and then leverages that use to increase revenue. 2 3 561. ByteDance has designed and aggressively marketed TikTok, the harmful and 4 addictive version of Douyin, to attract young Americans. 5 562. Like the other Defendants' products, TikTok depends on advertising revenue, which 6 has boomed. TikTok was projected to receive $11 billion in advertising revenue in 2022, over half 7 of which is expected to come from the United States.676 8 563. The initial iteration of TikTok allowed users to lip sync pop music by celebrities 9 who appealed primarily to teens and tweens (e.g., Selena Gomez and Ariana Grande). It labeled 10 folders with names attractive to youth (e.g., "Disney" and "school"); and included in those folders 11 songs such as "Can You Feel the Love Tonight" from the movie "The Lion King," "You've Got a 12 Friend in Me" from the movie "Toy Story," and other renditions covering school-related subjects 13 or school-themed television shows and movies.677 14 564. ByteDance also specifically and intentionally excluded videos that would not appeal 15 to young Americans, instructing TikTok moderators that videos of "senior people with too many 16 wrinkles" should not be permitted on users' "For You" pages because such content was "much less 17 attractive [and] not worth[] . . . recommend[ing]." 678 18 565. Even TikTok's sign-up process demonstrates that young users are what ByteDance 19 values most. In 2016, the birthdate for those signing up for the app defaulted to the year 2000 (i.e., 20 16 years old).679 21 676 Jessica Bursztynsky, TikTok says 1 billion people use the app each month, CNBC (Sept. 27, 22 2021), https://www.cnbc.com/2021/09/27/tiktok-reaches-1-billion-monthly-users.html; Bhanvi Staija, TikTok's ad revenue to surpass Twitter and Snapchat combined in 2022, Reuters (Apr. 11, 23 2022), https://www.reuters.com/technology/tiktoks-ad-revenue-surpass-twitter-snapchat- combined-2022-report-2022-04-11/. 24 677 Complaint for Civil Penalties, Permanent Injunction, and Other Equitable Relief ("Musical.ly Complaint") at p. 8, ¶¶ 26–27, United States v. Musical.ly, 2:19-cv-01439-ODW-RAO (C.D. Cal. 25 Feb. 27, 2019) Dkt. # 1. 678 Sam Biddle et al., Invisible Censorship: TikTok Told Moderators to Suppress Posts by "Ugly" 26 People and the Poor to Attract New Users, Intercept (Mar. 15, 2020), https://theintercept.com/2020/03/16/tiktok-app-moderatorsusers-discrimination/. 27 679 Melia Robinson, How to Use Musical.ly, The App With 150 million Users That Teens Are 28 Obsessed With, Bus. Insider (Dec. 7, 2016), https://www.businessinsider.com/how-to-use- musically-app-2016-12. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 175 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 182 of 278 1 3. ByteDance intentionally designed product features to addict children and adolescents. 2 3 566. TikTok's growth among young Americans has been further enabled by its defective 4 age verification and parental control procedures, which allow children under 13 unfettered access 5 to the app. 6 a. TikTok's age-verification measures are defective. 7 567. When a user first opens TikTok, they are prompted to "Login in to TikTok" or "Sign 8 up" for an account using a phone number or email address. TikTok then asks, "When's your 9 birthday?" 10 11 12 13 14 15 16 17 18 19 20 21 22 568. ByteDance does not verify the age that TikTok users report. Nor does it use any 23 method to verify that users who acknowledge they are minors have the consent of their parents or 24 legal guardians to use the product. In fact, at least as of 2020, TikTok still had not developed a 25 company position on age verification. 680 26 27 680 28 TIKTOK3047MDL-001-00060941 at *85 ("Minor Safety Policy & PnP,'" PowerPoint, January 2021). PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 176 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 183 of 278 1 569. ByteDance has designed TikTok so users can circumvent TikTok's age restrictions 2 by using TikTok without creating an account. TikTok allows users, no matter what age, to "browse 3 as [a] guest," and watch TikTok's "For You" page while TikTok's algorithm collects data about 4 that user and their viewing behavior. 681 5 570. ByteDance knows that many U.S. TikTok users under the age of 13 fail to report 6 their birth dates accurately.682 7 571. ByteDance's Trust and Safety team recognizes that one of the biggest challenges it 8 faces is "determining who is a minor (defined as users 13-17 years old)." 683 9 572. In 2019, the FTC acted on this admission and alleged that ByteDance failed to 10 comply with COPPA.684 11 573. TikTok settled the FTC claims, agreeing to a then-record civil COPPA penalty and 12 several forms of injunctive relief intended to protect children who use the product. 685 13 574. To comply with the terms of that settlement, ByteDance created "TikTok for 14 Younger Users," a "limited app experience" for users under the age of 13. 686 "TikTok for Younger 15 Users" does not permit users to "share their videos, comment on others' videos, message with users, 16 or maintain a profile or followers."687 However, users can still "experience what TikTok is at its 17 18 19 681 Browse as Guest, TikTok Support, https://support.tiktok.com/en/log-in-troubleshoot/log- 20 in/browse-as-guest. 682 Jon Russell, Musical.ly Defends its Handling of Young Users, As it Races Past 40M MAUs, 21 TechCrunch (Dec. 6, 2016), https://techcrunch.com/2016/12/06/musically-techcrunch-disrupt- london/. 22 683 TIKTOK3047MDL-001-00060811 at *16. 684 23 See Musical.ly Complaint, at p. 8, ¶¶ 26–27. 685 Natasha Singer, TikTok Broke Privacy Promises, Children's Groups Say, NY Times (May 14, 24 2020), https://www.nytimes.com/2020/05/14/technology/tiktok-kids- privacy.html#:~:text=TikTok%2C%20the%20popular%20app%20for%20making%20and%20sha 25 ring,20%20children%E2%80%99s%20and%20consumer%20groups%20said%20on%20Thursda y. 26 686 TikTok for Younger Users, TikTok (Dec. 13, 2019), https://newsroom.tiktok.com/en-us/tiktok- for-younger-users. 27 687 Dami Lee, TikTok Stops Young Users from Uploading Videos after FTC Settlement, Verge 28 (Feb. 27, 2019), https://www.theverge.com/2019/2/27/18243510/tiktok-age-young-user-videos- ftc-settlement-13-childrensprivacy-law. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 177 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 184 of 278 1 core" by recording and watching videos on TikTok. For that reason, experts state the app is 2 "designed to fuel [kids'] interest in the grown-up version." 688 3 575. Moreover, users under 13 can easily delete their age-restricted accounts and sign up 4 for an over-13 account on the same mobile device—without any restriction or verification—using 5 a fake birthdate. 6 576. The absence of effective age verification measures also means that adult users claim 7 to be children—with obvious dangers to the children on ByteDance's product. 8 b. TikTok's parental controls are defective. 9 577. Following the FTC settlement, ByteDance created a "Family Pairing" feature on 10 TikTok. The supposed purpose of that feature was to allow parents to link their accounts to their 11 children's accounts and enforce certain controls (such as screen time limits and restriction of 12 "content that may not be appropriate for all audiences"). 689 13 578. "Family Pairing" also is supposed to allow parents to prevent their children from 14 direct messaging other TikTok users. But ByteDance has designed TikTok's "Family Pairing" 15 feature so that it is not mandatory for minor users. And to use it, a parent or guardian is forced to 16 create their own TikTok account to pair it with their child's account. Further, the "Family Pairing" 17 feature is available only on the TikTok app. It provides no protection when a child accesses TikTok 18 through a web browser. Because this feature requires parents to know the name of their child's 19 account to pair it, youth can easily evade the protections of the "Family Pairing" feature by creating 20 anonymous accounts, again without parental approval or knowledge. 21 579. ByteDance further stymies parents' ability to supervise minors' use of TikTok by 22 permitting minor users to block their parent's profile, post ephemeral videos called "Stories" that 23 disappear after 24 hours, and post those stories to "Friends Only." 24 580. ByteDance could, but does not, adopt safety features that notify parents when minors 25 are engaging excessively with the product and are using it during sleeping hours. 26 688 Leonard Sax, Is TikTok Dangerous for Teens?, Inst. Fam. Stud. (Mar. 29, 2022), 27 https://ifstudies.org/blog/istiktok-dangerous-for-teens. 689 28 TikTok Introduces Family Pairing, TikTok Newsroom (April 15, 2020) https://newsroom.tiktok.com/en-us/tiktok-introduces-family-pairing. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 178 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 185 of 278 1 581. Until January 13, 2021, ByteDance interfered with parental supervision and 2 endangered children by defaulting all accounts, including those registered to children as young as 3 13, to "public." That allowed strangers to contact minor users regardless of age or location. 4 ByteDance also intentionally and actively promoted these types of connections by suggesting 5 accounts to follow through the "Find Friends" or "People You May Know" features. 6 582. Today, for users 16 and over, ByteDance still sets the default privacy setting for all 7 registered accounts to "public," meaning that anyone can view a user's profile, on or off TikTok, 8 request the user as a friend, or engage with the user's content. 690 9 c. ByteDance intentionally designed TikTok's defective features and algorithms to maximize engagement using automatic 10 content, time-limited experiences, intermittent variable rewards, reciprocity, and ephemeral content. 11 12 583. Like each of the other Defendants, ByteDance has designed and coded TikTok with 13 features that foster addictive and compulsive use by youth, leading to a cascade of additional mental 14 and physical injuries. 15 584. One of TikTok's defining features is its "For You" page (or "FYP"). According to 16 ByteDance, it is "central to the TikTok experience and where most of our users spend their time." 691 17 585. TikTok's FYP uses ByteDance's powerful machine-learning algorithms to select 18 content to feed users to maximize their engagement and thereby serve ByteDance's interests—as 19 opposed to simply responding to searches by users. As one industry commentator explained, 20 TikTok uses "a machine-learning system that analyzes each video and tracks user behavior so that 21 it can serve up a continually refined, never-ending stream of TikToks optimized to hold [users'] 22 23 24 25 26 690 See, e.g., Lauren E. Sherman et al., The Power of the Like in Adolescence: Effects of Peer Influence on Neural and Behavioral Responses to Social Media, 27(7) Psych. Sci. 1027–35 (July 27 2016), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5387999. 691 28 How TikTok recommends videos #ForYou, TikTok (June 18, 2020), https://newsroom.tiktok.com/en-us/howtiktok-recommends-videos-for-you. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 179 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 186 of 278 1 attention."692As another commentator put it, "you don't tell TikTok what you want to see. It tells 2 you."693 3 586. Zhu has remarked that, "[e]ven if you have tens of millions of users, you have to 4 keep them always engaged."694 Thus, according to Zhu, TikTok's algorithms are "focused primarily 5 on increasing the engagement of existing users." 695 6 587. An internal document titled "TikTok Algo 101," which TikTok has confirmed is 7 authentic, "explains frankly that in the pursuit of the company's 'ultimate goal' of adding daily 8 active users, it has chosen to optimize for two closely related metrics in the stream of videos it 9 serves: 'retention'—that is, whether a user comes back—and 'time spent.'" 696 10 588. "This system means that watch time is key," explained Guillaume Chaslot, the 11 founder of Algo Transparency.697 Chaslot noted that "rather than giving [people] what they really 12 want," TikTok's "algorithm tries to get people addicted[.]" 698 13 589. To fulfill this goal, the TikTok algorithm responds to a user's time spent watching 14 and engaging with a video by feeding them similar content. 699 As TikTok describes it, the 15 algorithms populate each user's FYP feed by "ranking videos based on a combination of factors" 16 that include, among others, any interests expressed when a user registers a new account, videos a 17 user likes, accounts they follow, hashtags, captions, sounds in a video they watch, certain device 18 692 19 Jia Tolentino, How TikTok Holds Our Attention, New Yorker (Sept. 30, 2019), https://www.newyorker.com/magazine/2019/09/30/how-tiktok-holds-our-attention. 20 693 Drew Harwell, How TikTok Ate the Internet, Wash. Post. (Oct. 14, 2022), https://www.theday.com/business/20221015/how-tiktok-ate-the-internet/. 21 694 Biz Carson, How a Failed Education Startup Turned Musical.ly, the Most Popular App You've Probably Never Heard Of, Business Insider (May 28, 2016), 22 https://www.businessinsider.com/what-is-musically-2016-5 (emphasis added). 695 23 Joseph Steinberg, Meet Musical.ly, the Video Social Network Quickly Capturing the Tween and Teen Markets, Inc. (June 2, 2016), https://www.inc.com/joseph-steinberg/meet-musically- 24 the-video-social-network-quicklycapturing-the-tween-and-teen-m.html. 696 Ben Smith, How TikTok Reads Your Mind, N.Y. Times (Dec. 5, 2021), 25 https://www.nytimes.com/2021/12/05/business/media/tiktok-algorithm.html. 697 Ben Smith, How TikTok Reads Your Mind, N.Y. Times (Dec. 5, 2021), 26 https://www.nytimes.com/2021/12/05/business/media/tiktok-algorithm.html. 698 Ben Smith, How TikTok Reads Your Mind, N.Y. Times (Dec. 5, 2021), 27 https://www.nytimes.com/2021/12/05/business/media/tiktok-algorithm.html. 699 28 Kaitlyn Tiffany, I'm Scared of the Person TikTok Thinks I Am, The Atlantic (June 21, 2021), https://www.theatlantic.com/technology/archive/2021/06/your-tiktok-feed-embarrassing/619257/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 180 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 187 of 278 1 settings, such as their language preferences and where they are located, and finally the likelihood 2 of the user's interest.700 3 590. ByteDance has designed TikTok's algorithm so that certain factors, such as time 4 spent watching a video, are more important to the algorithm than others. For example, TikTok 5 explains that, "whether a user finishes watching a longer video from beginning to end, would 6 receive greater weight than . . . whether the video's viewer and creator are both in the same 7 country."701 8 591. TikTok's algorithms are designed to begin working the minute a user opens the app. 9 The FYP shows the user a single, full-screen stream of videos, then records how the user reacts. "A 10 second of viewing or hesitation indicates interest; a swipe suggests a desire for something else." 702 11 With each data point collected, TikTok's algorithm winnows a mass of content to a single feed, 12 continually refined to keep users engaging often and at length. 13 592. This algorithmic encouragement of continuous scrolling and interaction makes it 14 hard for users to disengage from the app. A recent ByteDance-funded study, which imaged the 15 brains of TikTok and other social media product users, found that those using TikTok engaged with 16 the product about 10 times a minute, twice as often as with peer apps. 703 17 593. ByteDance leverages users' inability to disengage as a benefit to attract advertisers, 18 rather than taking steps to address the addictive nature of its product. A recent TikTok marketing 19 700 20 Investigation: How TikTok's Algorithm Figures Out Your Deepest Desires, Wall St. J. (Jul. 21, 2021), https://www.wsj.com/video/series/inside-tiktoks-highly-secretive-algorithm/investigation- 21 how-tiktok-algorithm-figures-out-your-deepest-desires/6C0C2040-FF25-4827-8528- 2BD6612E3796; see also How TikTok recommends videos #ForYou | TikTok Newsroom, 22 https://newsroom.tiktok.com/en-us/how-tiktok-recommends-videos-for-you. 701 Investigation: How TikTok's Algorithm Figures Out Your Deepest Desires, Wall St. J. (Jul. 21, 23 2021), https://www.wsj.com/video/series/inside-tiktoks-highly-secretive-algorithm/investigation- how-tiktok-algorithm-figures-out-your-deepest-desires/6C0C2040-FF25-4827-8528- 24 2BD6612E3796; see also How TikTok recommends videos #ForYou | TikTok Newsroom, https://newsroom.tiktok.com/en-us/how-tiktok-recommends-videos-for-you. 25 702 Investigation: How TikTok's Algorithm Figures Out Your Deepest Desires, Wall St. J. (Jul. 21, 2021), https://www.wsj.com/video/series/inside-tiktoks-highly-secretive-algorithm/investigation- 26 how-tiktok-algorithm-figures-out-your-deepest-desires/6C0C2040-FF25-4827-8528- 2BD6612E3796; see also How TikTok recommends videos #ForYou | TikTok Newsroom, 27 https://newsroom.tiktok.com/en-us/how-tiktok-recommends-videos-for-you. 703 28 TikTok Ads Break Through Better Than Tv and Drive Greater Audience Engagement, TikTok, https://www.tiktok.com/business/library/TikTokDrivesGreaterAudienceEngagement.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 181 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 188 of 278 1 document observed that "the TikTok audience is fully leaned in." 704 Marketing research 2 commissioned by TikTok found that compared to other social media sites, TikTok users evidenced 3 a higher frequency of rate per minute. TikTok boasted, "[o]ur algorithm and shorter video formats 4 create continuous cycles of engagement, making TikTok the leading platform for Information 5 Density."705 6 594. ByteDance also creates images and GIFs for users to incorporate into TikTok videos 7 to keep users returning to the product. And ByteDance has acquired publishing rights to thousands 8 of hours of music and video, which it provides its users to attach to the videos and pictures they 9 post on TikTok. 10 595. TikTok's powerful machine-learning algorithms dictate the content of each user's 11 FYP. An estimated 90-95% of the content viewed on TikTok comes from its algorithms (as opposed 12 to user selection), the highest among Defendants' products. 706 13 596. The algorithm encourages use of the product, regardless of whether that use is 14 enjoyable or healthy. From TikTok's perspective, it doesn't matter whether you're engaging with 15 a video because you're horrified or angry or upset—the engagement itself is the end goal. 16 597. As the algorithm continues to refine what users see, they are "more likely to 17 encounter harmful content."707 Indeed, TikTok's quest to monopolize user attention often forces 18 users down "rabbit holes" of harmful content. Users end up in these rabbit holes, and become 19 trapped in them, because TikTok has optimized its algorithm's design for retention and time spent 20 on the app.708 TikTok wants to keep users coming back as often as possible for as long as possible. 21 22 704 TikTok Ads Break Through Better Than Tv and Drive Greater Audience Engagement, TikTok, https://www.tiktok.com/business/library/TikTokDrivesGreaterAudienceEngagement.pdf. 23 705 TikTok Ads Break Through Better Than Tv and Drive Greater Audience Engagement, TikTok, 24 https://www.tiktok.com/business/library/TikTokDrivesGreaterAudienceEngagement.pdf. 706 Investigation: How TikTok's Algorithm Figures Out Your Deepest Desires, Wall St. J. (Jul. 21, 25 2021), https://www.wsj.com/video/series/inside-tiktoks-highly-secretive-algorithm/investigation- how-tiktok-algorithm-figures-out-your-deepest-desires/6C0C2040-FF25-4827-8528- 26 2BD6612E3796. 707 Inside TikTok's Algorithm: A WSJ Video Investigation, Wall St. J. (July 21, 2021), 27 https://www.wsj.com/articles/tiktok-algorithm-video-investigation-11626877477. 708 28 Ben Smith, How TikTok Reads Your Mind, N.Y. Times (Dec. 5, 2021), https://www.nytimes.com/2021/12/05/business/media/tiktok-algorithm.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 182 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 189 of 278 1 598. Once users are in a rabbit hole, it is extremely difficult to climb out. One user was 2 shown a few anti-vaccination conspiracy theory videos on his FYP and commented on them in an 3 attempt to refute the videos' claims. His feed was quickly overtaken with similar videos, and it took 4 him months of intentional interaction with the app to purge this content from his FYP. 709 In general, 5 escaping a rabbit hole requires a user to repeatedly and actively strategize ways to counter the 6 algorithm, pitting individual users' David against TikTok's machine-learning Goliath. 7 599. The Wall Street Journal documented the pernicious operation of ByteDance's 8 algorithms, as shown by a recent experiment. The experimenters used bots, each programmed with 9 various interests such as sports, forestry, dance, astrology, and animals. They did not disclose these 10 interests upon registration with TikTok. Instead, TikTok's algorithm quickly learned the assigned 11 interests from the bots' behavior—that is, "by rewatching or pausing on videos" related to the bot's 12 programmed interest.710 13 600. One bot watched 224 videos in 26 minutes, lingering over videos with hashtags for 14 "depression" or "sad." The algorithm quickly refined its output. Afterward, 93% of the videos 15 TikTok showed that bot were about depression or sadness. One post implored the bot to: "Just go. 16 Leave. Stop trying. Stop pretending. You know it and so do they. Do Everyone a favor and 17 leave."711 18 601. ByteDance's choices about how to design and structure its app—including choosing 19 not to implement effective age gating and parental controls, in addition to choosing to design 20 algorithms to maximize engagement through pushing extreme and outrageous content—go far 21 beyond benignly organizing the content of others. Instead, they create an environment and 22 experience suited to ByteDance's goal of maximizing ad revenues—an environment and 23 experience that is unreasonably dangerous to the children and teens ByteDance targets. 24 25 709 Kaitlyn Tiffany, I'm Scared of the Person TikTok Thinks I Am, The Atlantic (June 21, 2021), 26 https://www.theatlantic.com/technology/archive/2021/06/your-tiktok-feed-embarrassing/619257/. 710 Inside TikTok's Algorithm: A WSJ Video Investigation, Wall St. J. (July 21, 2021), 27 https://www.wsj.com/articles/tiktok-algorithm-video-investigation-11626877477. 711 28 Inside TikTok's Algorithm: A WSJ Video Investigation, Wall St. J. (July 21, 2021), https://www.wsj.com/articles/tiktok-algorithm-video-investigation-11626877477. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 183 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 190 of 278 1 602. In a follow-up experiment by the Wall Street Journal, bots were registered as users 2 between 13 and 15 years-old. One of those bots, programmed to pause on videos referencing drugs, 3 lingered briefly on "a video of a young woman walking through the woods with a caption" referring 4 to "stoner girls." The next day, the algorithm showed the bot a video about a "marijuana-themed 5 cake." Then, the "majority of the next thousand videos" that TikTok's algorithm produced "tout[ed] 6 drugs and drug use," including marijuana, psychedelics, and prescription drugs. 712 7 603. The algorithm immersed another bot—registered as a 13-year-old boy—into a rabbit 8 hole of videos related to bondage and sex, including videos explaining, among other things, "how 9 to tie knots for sex, recover from violent sex acts and discussing fantasies about rape." 713 The bot 10 simply searched for the term "onlyfans"—a site known for hosting adult entertainment—and 11 watched a handful of videos in the results before returning to the FYP. 714 The algorithm 12 subsequently bombarded the bot with videos about sex and, as the bot lingered on those videos, the 13 bot's feed became almost entirely dominated by sex-related videos. At one point, "more than 90 14 percent of [the] account's video feed was about bondage and sex." 715 15 604. The Wall Street Journal concluded "that through its powerful algorithms, TikTok 16 can quickly drive minors—among the biggest users of the app—into endless spools of content about 17 sex and drugs."716 In another follow-up experiment, the Wall Street Journal found that once 18 TikTok's algorithm determined that the bots would rewatch videos related to weight loss, it 19 "speedily began serving more, until weight-loss and fitness content made up more than half their 20 feeds—even if the bot never sought it out."717 21 712 Rob Barry et al., How TikTok Serves Up Sex and Drug Videos to Minors, Wall St. J. (Sept. 8, 22 2021), https://www.wsj.com/articles/tiktok-algorithm-sex-drugs-minors-11631052944. 713 23 Rob Barry et al., How TikTok Serves Up Sex and Drug Videos to Minors, Wall St. J. (Sept. 8, 2021), https://www.wsj.com/articles/tiktok-algorithm-sex-drugs-minors-11631052944. 714 24 Rob Barry et al., How TikTok Serves Up Sex and Drug Videos to Minors, Wall St. J. (Sept. 8, 2021), https://www.wsj.com/articles/tiktok-algorithm-sex-drugs-minors-11631052944. 25 715 Rob Barry et al., How TikTok Serves Up Sex and Drug Videos to Minors, Wall St. J. (Sept. 8, 2021), https://www.wsj.com/articles/tiktok-algorithm-sex-drugs-minors-11631052944. 26 716 Rob Barry et al., How TikTok Serves Up Sex and Drug Videos to Minors, Wall St. J. (Sept. 8, 2021), https://www.wsj.com/articles/tiktok-algorithm-sex-drugs-minors-11631052944. 27 717 Tawnell D. Hobbs, 'The Corpse Bride Diet': How TikTok Inundates Teens With Eating- 28 Disorder Videos, Wall St. J. (Dec. 17, 2021), https://www.wsj.com/articles/how-tiktok-inundates- teens-with-eating-disorder-videos-11639754848. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 184 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 191 of 278 1 605. Indeed, TikTok's algorithm recommended over 32,000 weight-loss videos over a 2 two-month period, "many promoting fasting, offering tips for quickly burning belly fat and pushing 3 weight-loss detox programs and participation in extreme weight-loss competitions." 718 4 606. Alyssa Moukheiber, a treatment center dietitian, explained that TikTok's algorithm 5 can push children into unhealthy behaviors or trigger a relapse of disordered eating. 719 Indeed, 6 several teenage girls interviewed by the Wall Street Journal reported developing eating disorders 7 or relapsing after being influenced by extreme diet videos TikTok promoted to them. 720 8 607. Their experiences are not unique. Katie Bell, a co-founder of the Healthy Teen 9 Project, explained that "the majority of her 17 teenage residential patients told her TikTok played 10 a role in their eating disorders."721 11 608. Others, like Stephanie Zerwas, an Associate Professor of Psychiatry at the 12 University of North Carolina at Chapel Hill, could not even recount how many of her young patients 13 told her that "I've started falling down this rabbit hole, or I got really into this or that influencer on 14 TikTok, and then it started to feel like eating-disorder behavior was normal, that everybody was 15 doing that."722 16 609. In December 2022, the Center for Countering Digital Hate ("CCDH") conducted a 17 similar study, creating TikTok accounts with a registered age of 13 in the United States, United 18 Kingdom, Canada, and Australia.723 For the first 30 minutes on the app, the accounts paused briefly 19 718 Tawnell D. Hobbs, 'The Corpse Bride Diet': How TikTok Inundates Teens With Eating- 20 Disorder Videos, Wall St. J. (Dec. 17, 2021), https://www.wsj.com/articles/how-tiktok-inundates- teens-with-eating-disorder-videos-11639754848. 21 719 Tawnell D. Hobbs, 'The Corpse Bride Diet': How TikTok Inundates Teens With Eating- 22 Disorder Videos, Wall St. J. (Dec. 17, 2021), https://www.wsj.com/articles/how-tiktok-inundates- teens-with-eating-disorder-videos-11639754848. 720 23 Tawnell D. Hobbs, 'The Corpse Bride Diet': How TikTok Inundates Teens With Eating- Disorder Videos, Wall St. J. (Dec. 17, 2021), https://www.wsj.com/articles/how-tiktok-inundates- 24 teens-with-eating-disorder-videos-11639754848. 721 Tawnell D. Hobbs, 'The Corpse Bride Diet': How TikTok Inundates Teens With Eating- 25 Disorder Videos, Wall St. J. (Dec. 17, 2021), https://www.wsj.com/articles/how-tiktok-inundates- teens-with-eating-disorder-videos-11639754848. 26 722 Tawnell D. Hobbs, 'The Corpse Bride Diet': How TikTok Inundates Teens With Eating- Disorder Videos, Wall St. J. (Dec. 17, 2021), https://www.wsj.com/articles/how-tiktok-inundates- 27 teens-with-eating-disorder-videos-11639754848. 723 28 Deadly by Design, Center for Countering Digital Hate (Dec. 2022), https://counterhate.com/wp-content/uploads/2022/12/CCDH-Deadly-by-Design_120922.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 185 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 192 of 278 1 on videos about body image and mental health and liked them. "Where researchers identified a 2 recommended video matching one of the below categories, they viewed the video for 10 seconds 3 and liked it. For all other videos, researchers would immediately scroll the For You feed to view 4 the next video recommended by TikTok."724 TikTok's algorithm seized on this information and 5 within minutes began recommending content about eating disorders and self-harm. 6 610. The CCDH report further illustrated TikTok's algorithms at work, noting that, for 7 an account that liked content about body image and mental health, the algorithm recommended 8 similar content every 39 seconds. As the 30 minutes went on, TikTok recommended more videos 9 related to eating disorders, suicide, and self-harm, as the graph below shows. 10 11 12 13 14 15 16 17 18 19 20 21 22 23 611. TikTok's rabbit holes are particularly problematic for young people, who lack the 24 necessary impulse control to stop watching. The more the user engages by viewing or hesitating on 25 a particular piece of content, the more TikTok's algorithms learn about the user. ByteDance uses 26 this feature to exploit the vulnerabilities of children and teenagers, and addict them to its product. 27 724 Tawnell D. Hobbs, 'The Corpse Bride Diet': How TikTok Inundates Teens With Eating- 28 Disorder Videos, Wall St. J. (Dec. 17, 2021), https://www.wsj.com/articles/how-tiktok-inundates- teens-with-eating-disorder-videos-11639754848. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 186 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 193 of 278 1 612. Indeed, ByteDance admits that its recommendation algorithm creates a "risk of 2 presenting an increasingly homogeneous stream of videos." 725 As the above-referenced studies and 3 experiments demonstrate, that homogeneous stream often includes harmful content, including posts 4 about depression, self-harm, drugs, and extreme diets. 5 613. Rather than preventing children from falling down these harmful rabbit holes or 6 encountering harmful content, ByteDance threw up its hands, insisting "[i]t is not TikTok's place 7 to decide for people what is or is not 'appropriate' for them or their teens." 726 8 614. ByteDance uses a series of interrelated design features that exploit known mental 9 processes to induce TikTok's users to use the product more frequently, for more extended periods, 10 and with more intensity (i.e., providing more comments and likes). ByteDance knows or should 11 have known that children, whose brains are still developing, are particularly susceptible to these 12 addictive features. 13 615. TikTok is defective in part because ByteDance designed the app so users cannot 14 disable the auto-play function on the FYP.727 As noted above, when a user opens the TikTok app 15 or visits the TikTok website, the product immediately begins playing a video on the user's FYP. 16 The user may request more videos with a simple upward swipe, and the product will deliver a 17 seemingly endless content stream. If a user does not proceed from a video, it continues to play on 18 an endless loop. The ability to scroll continuously induces a "flow-state" and distorts users' sense 19 of time. 20 616. The TikTok app interface is designed with only a limited number of buttons and 21 sections of the app for users to navigate, such that the design does not impede "flow." 22 617. The FYP also leverages principles of IVR to encourage compulsive usage, in the 23 same fashion as Instagram Reels. A user swipes to receive the next video, and each swipe offers 24 the prospect (but not the certainty) of dopamine-releasing stimuli. 25 26 725 How TikTok recommends videos #ForYou, TikTok (June 18, 2020), https://newsroom.tiktok.com/en-us/howtiktok-recommends-videos-for-you. 27 726 TIKTOK3047MDL-001-00060817. 727 28 2 Best Ways You Can Turn off TikTok Autoplay, Globe Calls (Dec. 16, 2022), https://globecalls.com/2-best-ways-you-can-turn-off-tiktok-autoplay/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 187 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 194 of 278 1 618. The cumulative effect of these features is addictive, compulsive engagement. As 2 researchers at the Brown University School of Public Health explained, "the infinite scroll and 3 variable reward pattern of TikTok likely increase the addictive quality of the app as they may induce 4 a flow-like state for users that is characterized by a high degree of focus and productivity at the task 5 at hand."728 6 619. Dr. Julie Albright, a Professor at the University of Southern California, similarly 7 explained that TikTok is so popular because users will "just be in this pleasurable dopamine state, 8 carried away. It's almost hypnotic, you'll keep watching and watching." Users "keep scrolling," 9 according to Dr. Albright, "because sometimes you see something you like, and sometimes you 10 don't. And that differentiation—very similar to a slot machine in Vegas—is key." 729 11 620. Aza Raskin, the engineer who designed infinite scroll, described the feature as being 12 "as if [social media companies are] taking behavioral cocaine and just sprinkling it all over your 13 interface, and that's the thing that keeps you coming back and back and back." Because the infinite 14 scroll does not "give your brain time to catch up with your impulses . . . you just keep scrolling." 730 15 621. To reinforce this addictive experience, ByteDance intentionally omits the concept 16 of time from their product, stripping information such as when a user uploaded a video from its 17 endless stream of content. In the FYP, there is no way to discern how long ago the video was posted, 18 or when the user who posted the video joined TikTok. 19 622. On at least some phones, TikTok is designed to cover the clock displayed at the top 20 of user's iPhones, preventing them from keeping track of the time spent on TikTok. 731 21 22 728 23 Sophia Petrillo, What Makes TikTok So Addictive? An Analysis of the Mechanisms Underlying the World's Latest Social Media Craze, Brown Undergraduate J. of Pub. Health (Dec. 13, 2021), 24 https://sites.brown.edu/publichealthjournal/2021/12/13/tiktok/. 729 John Koetsier, Digital Crack Cocaine: The Science Behind TikTok's Success, Forbes (Jan. 18, 25 2020), https://www.forbes.com/sites/johnkoetsier/2020/01/18/digital-crack-cocaine-the-science- behind-tiktoks-success/?sh=765d1b4178be. 26 730 John Koetsier, Digital Crack Cocaine: The Science Behind TikTok's Success, Forbes (Jan. 18, 2020), https://www.forbes.com/sites/johnkoetsier/2020/01/18/digital-crack-cocaine-the-science- 27 behind-tiktoks-success/?sh=765d1b4178be. 731 28 Louise Matsakis, On TikTok, There is No Time, Wired (October 3, 2019), https://www.wired.com/story/tiktok-time/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 188 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 195 of 278 1 623. ByteDance has designed the app so that users can see, however, how many times a 2 video was liked, commented on, or shared. So, the only thing users can quantify within the app is 3 the approval or disapproval of others. 4 624. In June 2022, after receiving public criticism regarding its product's effects on 5 people's mental health, ByteDance introduced various tools to purportedly encourage users to take 6 a break from infinite scrolling, such as a "Take a Break" reminder and time-limit caps. ByteDance 7 could but does not activate these tools by default. Even for minors, once they have exceeded 100 8 minutes of usage a day, TikTok only "reminds" them that these "Take a Break" tools exist upon 9 opening the app, but does not automatically activate them by default. 10 625. In addition to the defective infinite scroll, ByteDance has designed TikTok so it has 11 other design features that exploit social psychological impulses to induce children to use TikTok 12 daily and for extended periods of time, adding to the product's addictive nature. 13 626. Several TikTok features actively encourage users to generate ephemeral photos and 14 videos. This defect promotes compulsive use, because users risk missing the content posted by their 15 friends and others if they do not check TikTok at least daily. 16 627. A TikTok user can, for example, post expiring "Stories," short videos that disappear 17 after 24 hours. These videos do not otherwise appear in a user's feed. TikTok's live stream feature 18 is similar.732 19 628. A relatively new feature, "TikTok Now," pushes daily notifications to users to share 20 "authentic, real-time images or 10-second videos at the same time as your friends." 733 ByteDance 21 designed this feature so that once a user gets the notification, the user has three minutes to post an 22 image or video. That user cannot view friends' "TikTok Now" posts without sharing one of their 23 own, and posts submitted outside of the three-minute window are marked as "late." TikTok 24 preserves a user's history in a calendar view, adding to the pressure to visit the app daily and when 25 notified by TikTok to do so. ByteDance designed these defective features to increase 26 732 Hilary Anderson, Social media apps are 'deliberately addictive to users, BBC (July 4, 2018), 27 https://www.bbc.com/news/technology-44640959. 733 28 TikTok Now, TikTok, https://www.tiktok.com/creators/creator-portal/product-feature- updates/tiktok-now. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 189 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 196 of 278 1 responsiveness to notifications and keep young users locked into the product, as they do not want 2 to miss out on this perceived social activity. 3 629. Like "Snap Streaks," "TikTok Now" does not enhance the communication function 4 of the product, but simply exploits young users' susceptibility to persuasive design, teenage social 5 anxiety, and FOMO. ByteDance's insidious design of "TikTok Now" also employs point scoring 6 and competition with others to drive frequent and continuous engagement by children, who 7 otherwise risk checking in late and alienating other peers participating in the exchange. 8 630. Like the other Defendants' apps, ByteDance designed TikTok to leverage the 9 principle of IVR by encouraging users to like, share, or reshare videos that others have created or 10 posted. Receiving a "Like" or "Reshare" indicates that others approve of that user's content and 11 satisfies their natural, developmentally predictable desire for acceptance. As discussed above, 12 "Likes" activate the reward region of the brain and release dopamine to create a positive feedback 13 loop.734 Users return to TikTok again and again, hoping for yet another pleasurable experience. 735 14 631. ByteDance also designed TikTok to use reciprocity to manipulate users into using 15 the app. One example is the "Duet" feature, which allows users to post a video side-by-side with a 16 video from another TikTok user. Users utilize "Duet" to react to the videos of TikTok content 17 creators. ByteDance intends the response to engender a reciprocal response from the creator of the 18 original video, inducing them to return to the app. 19 632. Another "core feature" of TikTok that ByteDance has pursued are "challenges," 20 which are campaigns that compel users to create and post in TikTok certain types of videos, such 21 as performing a dance routine or a dangerous prank. By fostering competition and the social 22 rewards of posting a challenge video, ByteDance incentivizes users to engage with the product 23 continuously. 24 734 Rasan Burhan & Jalal Moradzadeh, Neurotransmitter Dopamine (DA) and its Role in the 25 Development of Social Media Addiction, 11(7) J. Neurology & Neurophysiology 507 (2020), https://www.iomcworld.org/openaccess/neurotransmitter-dopamine-da-and-its-role-in-the- 26 development-of-social-media-addiction.pdf. 735 27 Rasan Burhan & Jalal Moradzadeh, Neurotransmitter Dopamine (DA) and its Role in the Development of Social Media Addiction, 11(7) J. Neurology & Neurophysiology 507 (2020), 28 https://www.iomcworld.org/openaccess/neurotransmitter-dopamine-da-and-its-role-in-the- development-of-social-media-addiction.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 190 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 197 of 278 1 633. Harmful and dangerous interactions are a foreseeable consequence of TikTok's 2 engagement-maximization design. For example, numerous minor users have injured themselves or 3 others participating in viral pranks to obtain rewards and increase the number of likes, views, and 4 followers. 5 634. One such viral prank, "the Benadryl challenge," features a user filming themselves 6 taking large quantities of Benadryl to cause hallucinations or induce an altered mental state. Other 7 similar viral challenges include the "NyQuil Challenge," in which young people are encouraged to 8 eat chicken cooked in NyQuil; the "Milk Crate Challenge," where adolescents climb atop a stack 9 of milk crates and jump off; the "Penny Challenge," where young users are encouraged to plug a 10 charger halfway into an outlet while holding a penny against the exposed prongs; and the "Blackout 11 Challenge" where youth are encouraged to make themselves faint by holding their breath and 12 constricting their chest muscles or restricting airflow with a ligature around their neck. 13 635. TikTok challenges have led to serious health complications, seizures, and death, 14 with at least 12 children in the United States dying from the TikTok Blackout Challenge alone. 736 15 636. As more users engage with a challenge video, the creator of the video is awarded 16 increased "Likes," views, and followers. TikTok's reward features thus incentivize this dangerous 17 behavior, resulting too often in physical harm or humiliation in the obsessive pursuit of social 18 significance.737 19 637. ByteDance encourages businesses to create challenges as a form of marketing, 20 explaining that challenges are "geared towards building awareness and engagement," and "research 21 shows that they can deliver strong results" and increased return on ad spending "at every stage of 22 the funnel."738 While ByteDance extolls the revenue potential from challenges, young users 23 736 Quinn Nguyen, Don't let your kids try these 9 dangerous TikTok trends! 24 https://cyberpurify.com/knowledge/9-dangerous-tiktok-trends/; Olivia Carville, TikTok's Viral Challenges Keep Luring Young Kids to Their Deaths, Bloomberg (Nov. 30, 2022) 25 https://www.bloomberg.com/news/features/2022-11-30/is-tiktok-responsible-if-kids-die-doing- dangerous-viral-challenges. 26 737 See TIKTOK3047MDL-001-00000813 ("The algorithm is designed to surface viral content, regardless of its source."). 27 738 Branded Hashtag Challenge: Harness the Power of Participation, TikTok for Business (Mar. 28 16, 2022), https://www.tiktok.com/business/en-US/blog/branded-hashtag-challenge-harness-the- power-of-participation. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 191 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 198 of 278 1 continue to face new and serious harms as the challenges' stakes grow even more extreme and 2 dangerous. 3 d. ByteDance's defective features include impediments to discontinuing use. 4 5 638. Even if a user escapes the addictiveness of TikTok's design and decides to delete 6 their account, ByteDance makes doing so a lengthy and complex undertaking. The deletion process 7 is defectively designed to encourage users to retain their accounts, even if their stated reason for 8 deletion is that the product is endangering their safety or health. 9 639. When a user selects the "Deactivate or delete account" in the "Account" section of 10 the TikTok app, the user is presented an option: "Delete or deactivate?" Deactivating an account 11 will preserve the user's data, but hide it from the product; deleting, on the other hand, will 12 permanently delete all data associated with the account. 13 640. However, ByteDance designed TikTok so that deletion is not immediate. The data 14 and account are preserved for 30 days, during which time the user can reactivate their account. 15 641. If a user selects the "Delete account permanently" option, the user is asked "Why 16 are you leaving TikTok?" The user must select from the following list: (1) I'm leaving temporarily; 17 (2) I'm on TikTok too much; (3) Safety or privacy concerns; (4) Too many irrelevant ads; (5) 18 Trouble getting started; (6) I have multiple accounts; or (7) Another reason. 19 642. If a user selects "I'm on TikTok too much," ByteDance makes a last-ditch effort to 20 retain the user by reminding the user that a limit can be set on the user's watch time on the product. 21 If a user selects "Safety or privacy concerns," the user is provided a list of resources to "secure" 22 the account. If the user selects "Another reason," a written explanation must be provided. The only 23 option that does not provide or require further information is "I have multiple accounts." ByteDance 24 isn't worried about users deleting merely one account if they already have multiple others. 25 643. Once a user selects a reason for deletion, the next screen prompts the user to 26 download their TikTok data. 27 644. Before the user continues the deletion, the product requires the user to check a box 28 at the bottom of the screen that says, "[b]y continuing, you reviewed your data request and wish to PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 192 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 199 of 278 1 continue deleting your account." This contrasts with the process of a user "agreeing" to the Terms 2 of Service and Privacy Policy during the registration process, which does not require a separate 3 confirmation. 4 645. Once the user confirms a desire to continue with the deletion process, the product 5 takes the user to yet another screen, which yet again asks whether the user wants to "delete this 6 account?" The text also explains that the account will be deactivated for 30 days, during which the 7 user may reactivate the account, and after 30 days, the account and data associated with it will be 8 permanently deleted. It goes on to warn that if a user deletes the account, the user will no longer be 9 able to do many things in the app. 10 646. Once a user again confirms that they want to delete their account, TikTok requires 11 validation with a 6-digit code sent to the telephone number or email address associated with the 12 account. Only after the user receives and enters the code may they finally "delete" their account 13 (after waiting 30 days). 14 647. ByteDance's account deletion process is inadequate for children attempting to 15 escape its addictive and harmful product. Requiring a child to go through multiple steps, and 16 offering alternatives, as well as a list of things they are giving up, is designed to convince them to 17 change their mind. Moreover, requiring the user to maintain a deactivated account for 30 days, 18 rather than deleting it on demand, increases the chance that an addicted user will relapse and return 19 to the app. 20 648. ByteDance's intentionally cumbersome and defective deletion process prioritizes 21 the retention of young users, and ad revenue that they generate, over their well-being. 22 e. ByteDance's defective features inflict impossible image standards and encourage negative appearance comparison. 23 24 649. ByteDance designed TikTok with image-altering filters that harm users. These 25 filters allow children to artificially change their appearance, for example by lightening their skin 26 and eyes, giving them glowing tan skin, or giving them larger lips or fluttering eyelashes. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 193 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 200 of 278 1 650. Young people often then compare the filtered images to their real-life appearance, 2 developing a negative self-image based on unrealistic, artificial images. 739 Many young girls use 3 image-altering filters every day, harming their mental health. And those filters subconsciously 4 make girls feel imperfect and ugly, "reduc[ing] their self-compassion and tolerance for their own 5 physical flaws."740 6 651. So compelling is the desire to resemble more closely the filtered ideal that there are 7 online tutorials explaining how to recreate certain filters using makeup. 8 9 10 11 12 13 14 15 16 17 18 652. Children's idealization of their filtered image is externally reinforced when the 19 filtered images receive more likes, comments, and other interaction. Young people also compare 20 these interaction "scores" to those of friends and celebrities who use filters, reinforcing the idea 21 that beauty depends on matching a digital ideal. 22 653. But filters, retouch, and other editing tools available on TikTok often alter specific 23 facial features, such as the shape of a person's eyes and lips, in ways that would require medical 24 739 Anna Haines, From 'Instagram Face' To 'Snapchat Dysmorphia': How Beauty Filters Are 25 Changing The Way We See Ourselves, Forbes (Apr. 27, 2021 at 1:19 PM EDT), https://www.forbes.com/sites/annahaines/2021/04/27/from-instagram-face-to-snapchat- 26 dysmorphia-how-beauty-filters-are-changing-the-way-we-see-ourselves/?sh=3c32eb144eff. 740 27 Anna Haines, From 'Instagram Face' To 'Snapchat Dysmorphia': How Beauty Filters Are Changing The Way We See Ourselves, Forbes (Apr. 27, 2021 at 1:19 PM EDT), 28 https://www.forbes.com/sites/annahaines/2021/04/27/from-instagram-face-to-snapchat- dysmorphia-how-beauty-filters-are-changing-the-way-we-see-ourselves/?sh=3c32eb144eff. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 194 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 201 of 278 1 intervention to alter in real life. Children, particularly girls, are thus striving for a standard of beauty 2 that is functionally impossible to achieve, with every TikTok filter creating a test that they are 3 doomed to fail. 4 4. ByteDance facilitates the spread of CSAM and child exploitation. 5 654. ByteDance has designed various TikTok features that promote and dramatically 6 exacerbate sexual exploitation, the spread of CSAM, sextortion, and other socially maladaptive 7 behavior that harms children. 8 655. TikTok's design features enable the spread of this illegal material, and it receives 9 value in the form of increased user activity for disseminating these materials on the product. 10 656. TikTok allows users to add a location to publicly shared videos of themselves. 741 11 TikTok encourages the use of location services, "prompt[ing] [users] to turn on Location Services 12 when [users] browse the For You feed." 13 657. By providing access to a child user's present physical location, ByteDance 14 encourages predators to locate nearby children for purposes of sexual exploitation, sextortion, and 15 CSAM. 16 658. ByteDance designed TikTok with a "Your Private Videos," feature, where users can 17 create and store private videos that are only visible to the user, better known as "Post-in-Private" 18 accounts, where adult predators store, create, post, and share CSAM. Within days of following a 19 small number of "Post-in-Private" accounts, TikTok's algorithm begins recommending dozens of 20 other "Post-in-Private" accounts to follow, making it easy for predators to view and share even 21 more CSAM.742 22 23 24 25 26 741 Location Information on TikTok, TikTok, https://support.tiktok.com/en/account-and- 27 privacy/account-privacy-settings/location-services-on-tiktok. 742 28 Location Information on TikTok, TikTok, https://support.tiktok.com/en/account-and- privacy/account-privacy-settings/location-services-on-tiktok. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 195 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 202 of 278 1 659. These accounts are nominally private, but users can share their usernames and 2 passwords with other users to access these private videos.743 While ByteDance's user policy forbids 3 sharing passwords with other users, TikTok's design means that it is nonetheless very easy to do. 744 4 660. ByteDance designed TikTok to offer two-factor authentication but does not require 5 users to enable it. In fact, when a user creates a new account, the default setting disables the two- 6 factor authentication.745 7 661. Furthermore, TikTok allows more than one device to be simultaneously logged into 8 a single account, allowing multiple predators to use one "Post-in-Private" account simultaneously. 9 662. ByteDance's "Post-in-Private" accounts feature also facilitate the grooming of 10 children and adolescents by adult predators. Adult predators can store CSAM videos in "Your 11 Private Videos" and then show them to adolescent users as a grooming tool. Should adult predators 12 convince adolescent users to create CSAM of themselves in the "Post-in-Private" accounts, the 13 "Your Private Videos" feature makes it easy for the videos to be produced, uploaded, and stored. 14 663. Another defective feature of TikTok is its livestream product, "TikTok LIVE." 15 Although ByteDance's policy restricts access for anyone under eighteen to "TikTok LIVE," 16 TikTok's design, as discussed above, does not incorporate an age verification protocol, so it is easy 17 for underage users to access this feature. 746 18 664. Within "TikTok LIVE" is another feature called "LIVE Gifts" for "viewers to react 19 and show their appreciation for [] LIVE content in real-time. 747 TikTok then awards "Diamonds" 20 to LIVE creators based on the popularity of their content. "One way for creators to collect 21 22 743 23 Gracelynn Wan, These TikTok Accounts Are Hiding Child Sexual Abuse Material In Plain Sight, Forbes (Nov. 14, 2022) https://www.forbes.com/sites/alexandralevine/2022/11/11/tiktok- 24 private-csam-child-sexual-abuse-material/?sh=749d6cb63ad9. 744 TikTok Terms of Service, https://www.tiktok.com/legal/page/us/terms-of-service/en. 25 745 How your email and phone number are used on TikTok, TikTok, https://support.tiktok.com/en/account-and-privacy/personalized-ads-and-data/how-your-phone- 26 number-is-used-on-tiktok. 746 What is TikTok LIVE?, TikTok, https://support.tiktok.com/en/live-gifts-wallet/tiktok- 27 live/what-is-tiktok-live. 747 28 LIVE Gifts on TikTok, TikTok, https://support.tiktok.com/en/live-gifts-wallet/tiktok-live/live- gifts-on-tiktok. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 196 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 203 of 278 1 "Diamonds is to receive Gifts from viewers on [their] LIVE videos." Creators awarded "Diamonds" 2 "may obtain a Reward Payment in money or in virtual items." 748 3 665. ByteDance's design of the "LIVE Gifts" and "Diamonds" rewards greatly increases 4 the risk of adult predators targeting adolescent users for sexual exploitation, sextortion, and CSAM. 5 According to Leah Plunket, an assistant dean at Harvard Law School, "TikTok LIVE" is "the digital 6 equivalent of going down the street to a strip club filled with 15-year-olds." 749 "Livestreams on 7 [TikTok] are a popular place for men to lurk and for young girls—enticed by money and gifts—to 8 perform sexually suggestive acts." 750 9 666. Another of TikTok's defective features enables predators to communicate privately 10 with youth, with virtually no evidence of what was exchanged. The private messaging or "Direct 11 messaging" feature allows a user to send a direct private message to another user. Predators use 12 these messages to identify children willing to respond to a stranger's message and then prey on the 13 child's vulnerabilities. 14 667. Although Tiktok's features enable predators, TikTok does not have any feature to 15 allow users to specifically report CSAM.751 16 668. Federal law mandates that ByteDance reports suspected CSAM to NCEMC under 17 18 U.S.C. § 2258A. To limit and avoid its reporting requirements under federal law, ByteDance 18 purposely designed its products—which it knows are used by children, including children under 19 13—not to incorporate modern CSAM detection technology. This technology would be free for 20 ByteDance to implement within its product design. 21 22 748 23 LIVE Gifts on TikTok, TikTok, https://support.tiktok.com/en/live-gifts-wallet/tiktok-live/live- gifts-on-tiktok. 24 749 Alexandra Levine, How TikTok Live Became a Strip Club Filled with 15 Year Olds, Forbes (Apr. 27, 2022), https://www.forbes.com/sites/alexandralevine/2022/04/27/how-tiktok-live- 25 became-a-strip-club-filled-with-15-year-olds/?sh=5d6cf08d62d7. 750 Alexandra S. Levine, How TikTok LIVE Became 'A Strip Club Filled with 15-Year Olds,' 26 Forbes, https://www.forbes.com/sites/alexandralevine/2022/04/27/how-tiktok-live-became-a- strip-club-filled-with-15-year-olds/?sh=64c0447362d7. 27 751 Canadian Centre for Child Protection, Reviewing Child Sexual Abuse Material Reporting 28 Functions on Popular Platforms, https://protectchildren.ca/pdfs/C3P_ReviewingCSAMMaterialReporting_en.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 197 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 204 of 278 1 669. Furthermore, in violation of 18 U.S.C. § 2258A, ByteDance knowingly fails to 2 report massive amounts of material in violation of 18 U.S.C. § 2256 and 18 U.S.C. § 1466A. 3 670. ByteDance knowingly fails to take feasible, adequate, and readily available 4 measures to remove these contraband materials from its product in a timely fashion. 5 671. ByteDance made approximately 596 reports to NCMEC in 2019 and 22,692 reports 6 in 2020.752 However, ByteDance failed to report materials, including materials depicting Plaintiffs, 7 violating the reporting requirements of 18 U.S.C. § 2258A in 2019. 8 672. Users have reported "Post-in-Private" CSAM videos to TikTok, and ByteDance 9 responded that no violations of its policy were found. One user searched for and contacted multiple 10 TikTok employees to sound the alarm that CSAM was being created and shared within TikTok's 11 "Post-in-Private" accounts. This user did not receive a single response to her concerns. 753 12 673. ByteDance nonetheless continues to make false representations that they will "take 13 immediate action to remove content, terminate accounts, and report cases to NCMEC and law 14 enforcement as appropriate."754 15 674. ByteDance gains revenue for every daily user on TikTok in North America. Each 16 user and their data are worth income, and ByteDance continues to benefit financially from predators 17 who commit sexual abuse against children and/or share CSAM using ByteDance's product. 18 5. ByteDance failed to adequately warn Plaintiffs about the harms its product causes or to provide instructions regarding safe use. 19 20 675. Since TikTok's inception, ByteDance has failed to adequately warn young users 21 about the physical and mental health risks its product poses. These risks include, but are not limited 22 to, product abuse and addiction, sexual exploitation from adult users, dissociative behavior, damage 23 to body image, social isolation, and a plethora of mental health disorders like body dysmorphia, 24 25 752 Community guidelines enforcement report, TikTok (2022), https://www.tiktok.com/transparency/en-us/community-guidelines-enforcement-2020-2/. 26 753 Gracelynn Wan, These TikTok Accounts Are Hiding Child Sexual Abuse Material In Plain Sight, Forbes (Nov. 14, 2022), https://www.forbes.com/sites/alexandralevine/2022/11/11/tiktok- 27 private-csam-child-sexual-abuse-material/?sh=290dbfa63ad9 754 28 Protecting Against Exploitative Content, TikTok, https://newsroom.tiktok.com/en- us/protecting-against-exploitative-content. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 198 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 205 of 278 1 eating disorders, anxiety, depression, insomnia, ADD/ADHD exacerbation, suicidal ideation, self- 2 harm, suicide, and death. 3 676. ByteDance targets young users via advertising and marketing materials distributed 4 throughout traditional as well as digital media, including other social media products. ByteDance 5 fails to provide adequate warnings in advertising and marketing campaigns to potential adolescent 6 consumers of the physical and mental harms associated with using TikTok. 7 677. ByteDance heavily advertises its product on YouTube and Snapchat, where it knows 8 it can effectively reach younger users. In 2019, for example, 80 percent of TikTok's advertising 9 spending was on Snapchat.755 10 678. One TikTok ad compiles viral videos featuring people of all ages and sets the video 11 to the pandemic musical hit "Bored in the House," by a popular TikTok creator. The 15-second 12 video, titled "It Starts On TikTok," notes, "if it's in culture, it starts on TikTok." 756 Zhu highlighted 13 the importance of the U.S. teen market to TikTok, admitting that in China, "teenage culture doesn't 14 exist" because "teens are super busy in school studying for tests, so they don't have the time and 15 luxury to play social media apps." On the other hand, teen culture in the United States is "a golden 16 audience."757 17 679. Other advertisements ByteDance places on YouTube promote TikTok as a family- 18 friendly product. For example, one commercial features parents impersonating their children, 19 explaining that "parents roasting their kids is the best kind of family bonding." 758 Another TikTok 20 ad asks content creators what TikTok means to them. Responses include "family," "sharing special 21 22 23 755 24 TikTok – Snapchat's Biggest Advertiser – What's the Strategy, Media Radar (Feb. 24, 2020), https://mediaradar.com/blog/tiktok-snapchat-advertising-strategy/. 25 756 TikTok, It Starts on TikTok: Bored in the House, YouTube (Sept. 9, 2020), https://www.youtube.com/watch?v=DWZCgkmcIjE. 26 757 Paul Mozur, Chinese Tech Firms Forced to Choose Market: Home or Everywhere Else, N.Y. Times (Aug. 9, 2016), https://www.nytimes.com/2016/08/10/technology/china-homegrown- 27 internet-companies-rest-of-the-world.html. 758 28 Family Impressions, Compilation, TikTok's Official YouTube Page, https://www.youtube.com/watch?v=6EYzm25gW-s. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 199 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 206 of 278 1 moments with my daughter," and a featured appearance by well-known TikTok creator Addison 2 Rae, who says TikTok represents "family and fun." 759 3 680. ByteDance released another TikTok ad, part of the "It Starts on TikTok" ad 4 campaign, and scheduled it to release on the linear TV, digital media, digital out-of-home, radio 5 and TikTok's own social channels. 760 The tagline for the campaign was "[l]oving all of you and the 6 things you do. Celebrating you" and featured a series of viral clips of various cheerful scenes 7 depicting people gathered with friends and family of ages. 8 681. ByteDance is also one of the biggest advertisers on Snapchat. In 2019, ByteDance 9 accounted for 4.4% of Snapchat's advertising revenue. 761 ByteDance knows that advertising on 10 Snapchat is an effective way to reach a young audience. Snap claims that its Snapchat product 11 reaches 90% of people aged 13-24 years old, and 75% of 13-34 year olds in the United States. 12 682. Despite its funny, cheerful ads featuring smiling families and funny images, TikTok, 13 as designed, presents serious risks to young users on the platform, through its distinctive and 14 manipulative product features, including a lack of adequate age and identity verification tools, as 15 well as inadequate parental controls. 16 683. ByteDance fails to adequately warn young users of these risks beginning with the 17 first stages of the product registration process. At account setup, TikTok contains no warning labels, 18 banners, or conspicuous messaging to adequately inform adolescent users of product risks, potential 19 dangers, and physical and mental harm associated with usage of the product. Instead, ByteDance 20 allows underage users to easily create an account (or multiple accounts) and fully access the 21 product. 22 23 24 759 TikTok Creators Share Their Thoughts About TikTok, TikTok's Official YouTube Page 25 https://www.youtube.com/watch?v=KAvEGBv7HVM. 760 Todd Spangler, TikTok Launches Biggest-Ever Ad Campaign as Its Fate Remains Cloudy, 26 Variety (Aug. 10, 2020), https://variety.com/2020/digital/news/tiktok-advertising-brand- campaign-sale-bytedance-1234738607/. 27 761 Robert Williams, TikTok is the biggest advertiser on Snapchat, study says, MarketingDive 28 (March 16, 2020), https://www.marketingdive.com/news/tiktok-is-the-biggest-advertiser-on- snapchat-study-says/574164/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 200 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 207 of 278 1 684. ByteDance's lack of appropriate warnings continues once a child has TikTok. 2 ByteDance does not suitably inform child users that their data will be tracked, used to help build a 3 unique algorithmic profile, and potentially sold to TikTok's advertising clients. 4 685. Alarmingly, ByteDance also does not adequately warn young users before 5 facilitating adult connections and interactions that adult predators use its product. 6 686. ByteDance's failure to adequately warn young users about the risks of the product 7 continues even if they display signs of addiction or habitual and compulsive use. Besides the 8 disabled by default "Take a Break" reminder, ByteDance does not warn users when their screen 9 time reaches harmful levels or when young users are accessing the product on a habitual basis. 10 687. Not only does ByteDance fail to adequately warn users about the risks associated 11 with TikTok, but it also does not provide sufficient instructions on how children can safely use the 12 product. A reasonable and responsible company would instruct children on best practices and safety 13 protocols when using a product known to contain danger and health risks. 14 688. ByteDance, however, fails to adequately warn users that: 15 sexual predators use its product to produce and distribute CSAM; 16 adult predators targeting children for sexual exploitation, sextortion, and 17 CSAM are prevalent on ByteDance's product; 18 usage of its product can increase the risk of children being targeted and 19 sexually exploited by adult predators; 20 usage of its product can increase risky and uninhibited behavior in children, 21 making them easier targets to adult predators for sexual exploitation, 22 sextortion, and CSAM; and, 23 end-to-end encryption and/or the ephemeral nature of ByteDance's direct 24 messaging product prevents the reporting of CSAM. 25 689. ByteDance failed to adequately warn parents about all of the foregoing dangers and 26 harms. ByteDance's failure to adequately warn and instruct as set forth herein has proximately 27 caused significant harm to Plaintiffs' mental and physical well-being, and other injuries and harms 28 as set forth herein. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 201 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 208 of 278 1 689A. In addition, pursuant to Federal Rule of Civil Procedure 10(c), Plaintiffs adopt by 2 reference the following paragraphs of the Complaint of the Division of Consumer Protection of the 3 State of Utah v. TikTok, Inc., No. 230907634, pending in the Third Judicial District for Salt Lake 4 County, Utah (attached hereto as Exhibit C): 5, 26, 31-32, 37, 40-44, 49, 52, 79, 90-111, 115, 145- 5 148. 6 E. FACTUAL ALLEGATIONS AS TO GOOGLE 7 690. Eric Schmidt, the former CEO of Google and more recently, Alphabet, YouTube's 8 corporate parent, recently acknowledged the powerful, and purposeful, addictive effect of social 9 media. Social media products are about "maximizing revenue," Mr. Schmidt said, and the best way 10 to maximize revenue is to "maximize engagement." As Mr. Schmidt continued, in pursuit of their 11 goal of maximizing engagement to increase revenues, social media products "play into the 12 addiction capabilities of every human."762 13 691. Google's YouTube product is no exception. It includes specific, carefully calibrated 14 features that are known to exploit the mental processes of its users to keep them engaged for as 15 long, as frequently, and as intensely as possible. Google knows that children and teenagers who 16 flock in droves to its YouTube product are particularly susceptible to these features. The impact of 17 YouTube's addictive power on American youth has been devastating. 18 1. Background and overview of YouTube. 19 692. YouTube is a social media product that allows users to post and consume countless 20 hours of video content about virtually any topic imaginable. YouTube is available without any age 21 verification feature or adequate parental controls, and comes pre-installed in many Smart-TVs, 22 mobile devices, various digital media players like Roku, and video game consoles like PlayStation, 23 Wii, X-box and Nintendo. 24 693. YouTube allows users to search for specific video content. It also employs a 25 powerful algorithm that exploits detailed user information to target each individual user with hours 26 upon hours of videos recommended by YouTube. 27 762 28 Issie Lapowsky, Eric Schmidt: Social Media Companies 'Maximize Outrage' for Revenue, Protocol (Jan. 6, 2022), https://www.protocol.com/bulletins/eric-schmidt-youtube-criticism. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 202 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 209 of 278 1 694. A group of design experts and computer scientists created YouTube and launched 2 the product for public use in December 2005. 3 695. Technology behemoth Google quickly recognized YouTube's huge profit potential. 4 In 2006, just a year after YouTube's launch, Google acquired YouTube for more than $1.65 billion 5 in Google stock. At the time, Google's acquisition of YouTube was one of the largest-ever tech 6 acquisitions. 7 696. YouTube primarily generates revenue by selling advertising. The more people who 8 use YouTube and spend time on the site, the more ads YouTube can sell. 763 The ads are then 9 embedded or placed within the endless stream of videos recommended to the user by YouTube's 10 algorithm. 11 697. By 2012, YouTube users were watching close to four billion hours of video every 12 month. Yet, the average YouTube user spent just fifteen minutes daily engaged with the product. 764 13 Users "were coming to YouTube when they knew what they were coming to look for." 765 They 14 employed the product to identify and watch certain video content, and then they were done. 15 698. To drive greater revenue, "YouTube . . . set a company-wide objective to reach one 16 billion hours of viewing a day[.]"766 17 699. As Susan Wojcicki, YouTube's CEO explained, the goal of a "billion hours of daily 18 watch time gave our tech people a North Star." 767 19 20 21 763 Mark Bergen, YouTube Executives Ignored Warnings, Letting Toxic Videos Run Rampant, 22 Bloomberg (Apr. 2, 2019), https://www.bloomberg.com/news/features/2019-04-02/youtube- executives-ignored-warnings-letting-toxic-videos-run-rampant. 764 23 John Seabrook, Streaming Dreams: YouTube Turns Pro, New Yorker (Jan. 16, 2012), https://www.newyorker.com/magazine/2012/01/16/streaming-dreams. 24 765 Casey Newton, How YouTube Perfected the Feed, Verge (Aug. 30, 2017), https://www.theverge.com/2017/8/30/16222850/youtube-google-brain-algorithm-video- 25 recommendation-personalized-feed. 766 Mark Bergen, YouTube Executives Ignored Warnings, Letting Toxic Videos Run Rampant, 26 Bloomberg (Apr. 2, 2019), https://www.bloomberg.com/news/features/2019-04-02/youtube- executives-ignored-warnings-letting-toxic-videos-run-rampant. 27 767 Mark Bergen, YouTube Executives Ignored Warnings, Letting Toxic Videos Run Rampant, 28 Bloomberg (Apr. 2, 2019), https://www.bloomberg.com/news/features/2019-04-02/youtube- executives-ignored-warnings-letting-toxic-videos-run-rampant. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 203 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 210 of 278 1 700. Google decided that "the best way to keep eyes on the site" was to introduce a 2 feature that would "[recommend] videos, [that were playing] or after one was finished." 768 3 701. That new product feature uses a recommendation algorithm to identify and push 4 additional videos to users, which YouTube plays automatically, through a feature called "autoplay." 5 Autoplay begins the next video as soon as the previous videos ends, creating a constant stream of 6 content. 7 702. Google's design changes worked. Today, YouTube "has over 2 billion monthly 8 logged-in users."769 And that 2 billion figure does not capture all product usage because YouTube, 9 by design, allows users to consume videos without logging in or registering an account. 10 2. Google intentionally encourages youth to use YouTube and then leverages that use to increase revenue. 11 12 703. Google knows that children and teenagers use YouTube in greater proportions than 13 older demographics. YouTube now ranks as the world's most popular social media product for 14 minors. According to one recent report, more than 95% of children ages 13-17 have used 15 YouTube.770 Nearly 20% of U.S. teens use YouTube almost constantly. 771 Among U.S. teenagers 16 who regularly use social media, 32% "wouldn't want to live without" YouTube. 772 17 704. Rather than ensuring minors are not inappropriately or excessively using YouTube, 18 Google has sought to dominate their attention. 19 705. YouTube's age controls are defective (or non-existent, since registration is not 20 required). In addition, Google has developed and marketed a version of YouTube, YouTube Kids, 21 768 Mark Bergen, YouTube Executives Ignored Warnings, Letting Toxic Videos Run Rampant, 22 Bloomberg (Apr. 2, 2019), https://www.bloomberg.com/news/features/2019-04-02/youtube- executives-ignored-warnings-letting-toxic-videos-run-rampant. 769 23 YouTube for Press, YouTube, https://blog.youtube/press/. 770 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 24 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 2022. 25 771 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 26 2022. 772 27 Victoria Rideout et al., Common Sense Census: Media Use by Tweens and Teens, 2021 at 31, Common Sense Media (2022), 28 https://www.commonsensemedia.org/sites/default/files/research/report/8-18-census-integrated- report-final-web_0.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 204 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 211 of 278 1 explicitly targeted at children under 13. Google developed this product to encourage early—and 2 therefore lasting—adoption of YouTube by children. 3 706. Google knows that a robust and committed base of young users is key to maximizing 4 advertising revenue. Indeed, it has aggressively touted its hold on child users to advertisers. 5 707. In 2014, for example, Google pitched its YouTube product to Hasbro, a popular toy 6 manufacturer, and specifically boasted of the product's immense popularity among children, noting 7 that it was "unanimously voted as the favorite website of kids 2-12" and that "93% of tweens" use 8 the product.773 9 708. In 2015, Google gave a similar presentation to toy manufacturer Mattel, the maker 10 of Barbie and other popular kids' toys, highlighting children's widespread use of YouTube to 11 persuade Mattel to display digital ads on the site.774 12 709. The FTC has aptly summarized Google's pitch to advertisers concerning the value 13 of its youth user base.775 For example, Google boasted that YouTube "is today's leader in reaching 14 children age 6-11;" "the new 'Saturday Morning Cartoons';" "unanimously voted as the favorite 15 website of kids 2-12;" "the #1 website regularly visited by kids;" and used by "93% of tweens." 776 16 710. Many of YouTube's most-viewed videos are kid-focused, and the most subscribed 17 and highest paid YouTubers are children. With over 12 billion views, "Baby Shark Dance," a video 18 aimed at toddlers, is the most viewed video in the history of YouTube–and it and five other child- 19 focused videos make up the top ten YouTube videos of all time. 777 Child creators also dominate 20 773 Complaint for Permanent Injunction, Civil Penalties, and Other Equitable Relief, FTC v. 21 Google LLC et al., No. 1-19-cv-02642-BAH, at 6 (D.D.C. Sept. 4, 2019) Dkt. #1-1. https://www.ftc.gov/system/files/documents/cases/youtube_complaint_exhibits.pdf. 774 22 Complaint for Permanent Injunction, Civil Penalties, and Other Equitable Relief, FTC v. Google LLC et al., No. 1-19-cv-02642-BAH, at 3 (D.D.C. Sept. 4, 2019) Dkt. #1-1. 23 https://www.ftc.gov/system/files/documents/cases/youtube_complaint_exhibits.pdf. 775 Google and YouTube Will Pay Record $170 Million for Alleged Violations of Children's 24 Privacy Law, FTC (Sept. 4, 2019),https://www.ftc.gov/news-events/news/press- releases/2019/09/google-youtube-will-pay-record-170-million-alleged-violations-childrens- 25 privacy-law. ("YouTube touted its popularity with children to prospective corporate clients", said FTC Chairman Joe Simons.) 26 776 Complaint for Permanent Injunction, Civil Penalties, and Other Equitable Relief, FTC v. Google LLC et al., No. 1-19-cv-02642-BAH, at 3,12, and 6-7 (D.D.C. Sept. 4, 2019) Dkt. #1-1. 27 https://www.ftc.gov/system/files/documents/cases/youtube_complaint_exhibits.pdf. 777 28 Most Viewed Videos of All Time • (Over 700M views) - YouTube. https://www.youtube.com/playlist?list=PLirAqAtl_h2r5g8xGajEwdXd3x1sZh8hC PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 205 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 212 of 278 1 top-earner lists year after year. Ryan Kaji of Ryan's World (f/k/a Ryan ToysReview), a channel 2 featuring now 12-year-old Ryan Kaji unboxing children's toys, has been among YouTube's Top 3 10 most-subscribed channels in the United States since 2016. 778 Ryan started Ryan's World in 2015 4 when he was only 3. By 2017, his videos had over 8 billion views, and by 2018, he was the highest- 5 earning YouTuber in the world.779 6 711. As with other defendants, once Google lures children in, it then mines them (and all 7 other users) for a breathtaking amount of data. Google's current privacy policy, which includes the 8 YouTube product's data collection, reveals how sweeping this data collection is. It states that 9 Google tracks: 10 a. "information about the apps, browsers, and devices you use to access Google services . . . include[ing] unique 11 identifiers, browser type and settings, device type and settings, operating system, mobile network information 12 including carrier name and phone number, and application version number. We also collect information about the 13 interaction of your apps, browsers, and devices with our services, including IP address, crash reports, system activity, 14 and the date, time, and referrer URL of your request." 15 b. "your activity in our services . . . includ[ing] terms you search for[;] videos you watch[;] views and interactions 16 with content and ads[;] voice and audio information[;] purchase activity[;] people with whom you communicate or 17 share content[;] activity on third-party sites and apps that use our services[;] and Chrome browsing history you've 18 synced with your Google Account." 19 c. "Your location information [including] GPS and other sensor data from your device[;] IP address[;] activity on 20 Google services, such as your searches and places you label like home or work[;] [and] information about things near 21 22 778 Madeline Berg, The Highest-Paid YouTube Stars of 2019: The Kids Are Killing It, Forbes 23 (Dec. 18, 2019), https://www.forbes.com/sites/maddieberg/2019/12/18/the-highest-paid-youtube- stars-of-2019-the-kids-are-killing-it/?sh=4c3df9a438cd; Madeline Berg, The Highest-Paid 24 YouTube Stars 2017: Gamer DanTDM Takes The Crown With $16.5 Million, Forbes (Dec. 7, 2017), https://www.forbes.com/sites/maddieberg/2017/12/07/the-highest-paid-youtube-stars- 25 2017-gamer-dantdm-takes-the-crown-with-16-5-million/?sh=72de79413979 779 26 Gamer DanTDM Takes The Crown With $16.5 Million, Forbes (Dec. 7, 2017), https://www.forbes.com/sites/maddieberg/2017/12/07/the-highest-paid-youtube-stars-2017- 27 gamer-dantdm-takes-the-crown-with-16-5-million/?sh=72de79413979; Natalie Robehmed & Madeline Berg, Highest-Paid YouTube Stars 2018: Markiplier, Jake Paul, PewDiePie And More, 28 Forbes (Dec. 3, 2018), https://www.forbes.com/sites/natalierobehmed/2018/12/03/highest-paid- youtube-stars-2018-markiplier-jake-paul-pewdiepie-and-more/?sh=7d909c3f909a. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 206 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 213 of 278 1 your device, such as Wi-Fi access points, cell towers, and Bluetooth-enabled devices;"780 2 712. Google's privacy policy also indicates that, like other Defendants, it purchases data 3 about its users from data brokers, which it euphemistically refers to as "trusted partners" or 4 "marketing partners."781 5 713. As with other Defendants, YouTube's collection and analysis of user data allows it 6 to assemble virtual dossiers on its users, covering hundreds if not thousands of user-specific data 7 segments. This, in turn, allows advertisers to micro-target marketing and advertising dollars to very 8 specific categories of users, who can be segregated into pools or lists using YouTube's data 9 segments. Advertisers purchase ad real estate space on users' feeds, which allow them to place the 10 right ads in front of these micro-targeted segments of users--including children, both in the main 11 YouTube frame and in the YouTube Kids product. Only a fraction of these data segments come 12 from content knowingly designated by users for publication or explicitly provided by users in their 13 account profiles. Instead, many of these data segments are collected by YouTube through 14 surveillance of each user's activity while using the product and even when logged off the product. 782 15 714. As with Meta, Google's data policy does not inform users, and did not inform 16 Plaintiffs, that the more time individuals spend using YouTube, the more ads Google can deliver 17 and the more money it can make, or that the more time users spend on YouTube, the more YouTube 18 learns about them, and the more it can sell to advertisers the ability to micro-target highly 19 personalized ads. 20 715. Google's secret virtual dossiers on its users, including child users, fuel its 21 algorithms. The company relies on this data—including data plainly reflecting use by children—to 22 train its algorithms. A Google engineer explained in a 2014 presentation: 23 What do I mean by a training example? It's a single-user 24 experience. On YouTube, perhaps it's that one [Thomas the Tank Engine] webpage my son saw six months ago, along with all the 25 recommendations that we showed him. We also record the outcome to know whether the recommendations we made are good or 26 780 Information Google Collects. https://policies.google.com/privacy?hl=en#infocollect. 27 781 Information Google Collects. https://policies.google.com/privacy?hl=en#infocollect. 782 28 About Targeting for Video Campaigns, Google, https://support.google.com/youtube/answer/2454017?hl=en. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 207 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 214 of 278 1 whether they're bad. That's a single training exercise. On a large property, you can easily get into hundreds of billions of these. 783 2 The engineer illustrated this with a slide, excerpted below, presenting how algorithmic analysis 3 both structured the format of recommendations of Thomas the Tank Engine YouTube videos and 4 provided information to inform algorithmic training through user engagement: 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 716. Through these and other efforts, YouTube has delivered massive amounts of 21 advertising revenue to Google. In 2021 alone, YouTube generated about $29 billion in revenue 22 selling ads on its site.784 23 24 783 Alex Woodie, Inside Sibyl, Google's Massively Parallel Machine Learning Platform, 25 Datanami (Jul. 17, 2014) https://www.datanami.com/2014/07/17/inside-sibyl-googles-massively- parallel-machine-learning-platform/. 26 784 Andrew Hutchinson, YouTube Generated $28.8 Billion in Ad Revenue in 2021, Social Media 27 Today (Feb. 2, 2021), https://www.socialmediatoday.com/news/youtube-generated-288-billion- in-ad-revenue-in-2021-fueling-the-creator/618208/; Jennifer Elias, YouTube Is a Media 28 Juggernaut That Could Soon Equal Netflix in Revenue, CNBC (Apr. 27, 2021), https://www.cnbc.com/2021/04/27/youtube-could-soon-equal-netflix-in-revenue.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 208 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 215 of 278 1 3. Google intentionally designed product features to addict children and adolescents. 2 3 717. Google devised and continues to employ interrelated product features to increase 4 usage and maximize engagement by teenagers and children. Simply put, YouTube's product 5 features are engineered to induce excessive use and to addict adolescents and children to the 6 product. 7 a. Google's age-verification measures and parental controls are defective. 8 9 718. Google's strategy to entrench minor users begins with access. The company 10 purports to impose a minimum age requirement and claims to verify the age of its users. But those 11 features are defective, as they do little to prevent children and teenagers from using the product. 12 719. Anyone with access to the Internet, regardless of age, can use YouTube and access 13 every video available through the product without registering an account or verifying their age. 14 YouTube does not even ask for age information before allowing users to consume YouTube videos. 15 720. A user needs an account to post content or like (or comment) on videos. But to get 16 one, a user needs only enter a valid email address and a birthday. Google does nothing to verify the 17 birthday entered by users in the U.S.—and the product freely permits users to change their birthdays 18 in their account settings after creating an account. 19 721. YouTube's defective age verification feature means that Google fails to protect 20 children from other product features discussed below that Google knows to be harmful to kids. 21 722. For example, for users 13-17, Google claims to disable YouTube's autoplay feature. 22 However, that measure is virtually meaningless because children can use YouTube without logging 23 into any account or by logging in but misreporting their age. 24 723. Even if children use YouTube Kids, that product contains many of the same defects 25 YouTube does, including a harmful, manipulative algorithm, as alleged below. 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 209 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 216 of 278 1 724. Google cannot credibly claim that it is unaware of the fact and extent of youth usage 2 of YouTube. Google's system can "identify children as being much younger than 13." 785 According 3 to Tracking Exposed, YouTube can rapidly identify a user as a child. 786 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 725. Google engineers have publicly admitted YouTube's algorithm tracks user age. As 22 Google engineers outlined in a 2016 paper on YouTube's recommendation system, "[d]emographic 23 features are important for providing priors so that the recommendations behave reasonably for new 24 users. The user's geographic region and device are embedded and concatenated. Simple binary and 25 26 785 Tracking Exposed, Report: Non-Logged-In Children Using YouTube at 6 (Apr. 2022), 27 https://tracking.exposed/pdf/youtube-non-logged-kids-03July2022.pdf. 786 28 Tracking Exposed, Report: Non-Logged-In Children Using YouTube at 15, 18 (Apr. 2022), https://tracking.exposed/pdf/youtube-non-logged-kids-03July2022.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 210 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 217 of 278 1 continuous features such as the user's gender, logged-in state and age are input directly into the 2 network as real values normalized to [0; 1]." 787 3 726. The Tracking Exposed Report indicated that there was "strong evidence" that 4 Google's systems continue to refine and develop a more precise estimate for under 18 users, but 5 the product does not "redirect them to YouTube Kids."788 6 b. YouTube is defectively designed to inundate users with features that use intermittent variable rewards and reciprocity. 7 8 727. Google uses a series of interrelated design features that exploit known mental 9 processes to induce YouTube's users to use the product more frequently, for more extended periods, 10 and with more intensity (i.e., providing more comments and likes). Google knows children and 11 adolescents, whose brains are still developing, are particularly susceptible to these addictive 12 features. 13 728. Google designed its product so that when children and teenagers use it, they are 14 inundated with interface design features specifically designed to dominate their attention and 15 encourage excessive use. Every aspect of how YouTube presents the format of a given page with a 16 video is structured to ensure unimpeded viewing of the videos, alongside download, like, and share 17 buttons, plus recommendations for more videos to watch. The organization of these features is 18 carefully calibrated to adjust to the space constraints of a user's device, such that minimal effort is 19 needed to watch a video unimpeded. YouTube even has an ambient mode that uses dynamic color 20 sampling so that the YouTube product adapts to the video being watched and the user is not 21 distracted by the video's borders. 789 22 729. Like the other Defendants, Google has designed YouTube with features that exploit 23 neuropsychology to maximize the time users (including children) spend using the product. 24 25 787 Paul Covington et al., Deep Neural Networks for YouTube Recommendations, Google (2016), 26 https://storage.googleapis.com/pub-tools-public-publication-data/pdf/45530.pdf. 788 Tracking Exposed, Report: Non-Logged-In Children Using YouTube at 6, 19 (Apr. 2022), 27 https://tracking.exposed/pdf/youtube-non-logged-kids-03July2022.pdf. 789 28 YouTube rolling out black dark theme, 'Ambient Mode,' and other video player updates (Oct. 24, 2022). https://9to5google.com/2022/10/24/youtube-ambient-mode/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 211 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 218 of 278 1 730. IVR features, such as notifications and likes, compel YouTube content creators and 2 consumers, particularly children, to use the product habitually and excessively. For example, in 3 order to create and upload content to YouTube, a user under 13 may submit a fictitious birthdate in 4 order to gain access to posting privileges. Once the young user has a logged–in account, they are 5 capable of receiving notifications and likes. For example, the logged in user can subscribe to 6 various YouTube channels, which in turn will send them notifications from various channels they 7 follow. Similarly, young content creators who upload videos to YouTube are able to track the likes 8 received by the video. These features psychologically reward creators who upload videos to 9 YouTube. As explained above, receiving a "Like" shows others' approval and activates the brain's 10 reward region.790 Thus, users' ability to like content encourages creators to use the product 11 compulsively, seeking additional pleasurable experiences. 12 731. Another YouTube defect is the design Google engineers deploy to induce "flow" 13 state among users, which as described above is dangerous to children because it induces excessive 14 use and poses a risk of addiction, compulsive use, and sleep deprivation. 15 732. YouTube uses two design features that induce flow state. The first is its panel of 16 recommended videos. YouTube recommends videos both on the home page and on each video page 17 in the "Up Next" panel.791 This panel pushes an endless stream of videos that YouTube's algorithm 18 selects and "suggests" to keep users watching by teasing a pipeline of upcoming content. 19 733. The second feature is autoplay, which complements the Up Next panel and 20 seamlessly takes users through the list of upcoming videos without users having to affirmatively 21 click on or search for other videos. This constant video stream—comprised of videos recommended 22 by YouTube's algorithm—is the primary way Google increases the time users spend using its 23 product. This endless video succession induces users to enter a flow state of consumption, which 24 is particularly dangerous for children. 25 26 790 See, e.g., Lauren E. Sherman et al., The Power of the Like in Adolescence: Effects of Peer Influence on Neural and Behavioral Responses to Social Media, 27(7) Psych. Sci. 1027–35 (July 27 2016), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5387999/. 791 28 Recommended Videos, YouTube, https://www.youtube.com/howyoutubeworks/product- features/recommendations/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 212 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 219 of 278 1 734. In an April 2021 letter to YouTube CEO Susan Wojcicki, the House Committee on 2 Oversight and Reform criticized the autoplay feature: 3 This places the onus on the child to stop their viewing activity, rather than providing a natural break or end point. Without that natural 4 stopping point, children are likely to continue watching for long periods of time.792 5 735. This defect is particularly acute for Google's recently launched YouTube Shorts. 6 YouTube Shorts enables users to create short videos up to sixty seconds in length, in a full-screen 7 format popularized by TikTok and copied by Instagram Reels. As in Reels and TikTok, Shorts are 8 presented in an algorithmically generated feed; users can watch new videos by swiping up on their 9 smartphones. Instead of presenting videos chronologically, they are organized in a manner to drive 10 the most watch time, as dictated by the algorithm. Indeed, Google hired TikTok's North American 11 head, Kevin Ferguson, and other TikTok engineers to develop YouTube Shorts. 793 And much like 12 those other products, the ability to scroll continuously through YouTube Shorts content induces a 13 "flow-state," distorting users' sense of time and facilitating extended use. 14 736. An important target audience for YouTube Shorts is children. For example, 15 YouTube Shorts features content, such as child "influencers," that appeals to children. YouTube 16 Shorts also contains similar defects to other Defendants' short form products, including the ability 17 to scroll continuously through YouTube Shorts, inducing a "flow-state" that distorts users' sense 18 of time and facilitates extended use, and dangerous exploitation of "social comparison" techniques 19 by promoting misleadingly idealized portrayals from influencers and others who are rewarded for 20 posting popular material. 21 737. Almost immediately upon launch, Google began marketing YouTube Shorts to 22 children. For example, Google launched an advertisement featuring images of children and 23 teenagers (like in the screenshot below) engaging with the YouTube Shorts product. 24 25 792 Letter from Rep. Raja Krishnamoorthi, Chairman, Subcomm. on Economic and Consumer Policy, to Susan Wojcicki, CEO, YouTube (Apr. 6, 2021), 26 https://oversightdemocrats.house.gov/sites/democrats.oversight.house.gov/files/2021-04- 06.RK%20to%20Wojcicki-YouTube%20re%20YouTube%20Kids%20Content.pdf, 27 793 Richard Nieva, In the Age of TikTok, YouTube Shorts Is a Platform in Limbo, Forbes (Dec. 20, 28 2022), https://www.forbes.com/sites/richardnieva/2022/12/20/youtube-shorts-monetization- multiformat/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 213 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 220 of 278 1 2 3 4 5 6 7 8 9 10 738. Similarly, another advertisement for Shorts explains how creators on YouTube can 11 keep revenue generated by their Shorts viewership, while an image of a video creator young enough 12 to be in braces appears on screen.794 13 14 15 16 17 18 19 20 21 22 23 24 739. Shorts is one of YouTube's interrelated design features that exploit known mental 25 processes to induce YouTube users to use the product more frequently, for more extended periods, 26 and with more intensity (i.e., providing more comments and likes). Not surprisingly, given its 27 794 28 Made on YouTube: New ways to join YPP, Shorts Monetization & Creator Music. https://www.youtube.com/watch?v=h6TrvCV3NdU. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 214 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 221 of 278 1 copycat origin, the defects in Shorts replicate the defects in TikTok and Instagram Reels, discussed 2 above. Google knows or should have known that children, whose brains are still developing, are 3 particularly susceptible to such addictive features. 4 740. YouTube has monetized users' susceptibility to IVR by allowing creators who 5 obtain more than a thousand subscribers with four-thousand valid public watch hours to qualify for 6 the YouTube Partner Program. Once a creator obtains this elite status, they are rewarded with 7 "Super Chat" and "Super Stickers"—special images or distinct messages that other users can 8 purchase and place on a creator's channel. 795 Paid messages, including the amount donated, are 9 visible to all users. And the more a user pays for these promotions, the more prominent and longer 10 the image is displayed. Both features are intended to allow a user to show support for, or connect 11 with, their favorite YouTube creators. Similar to the "Likes" feature, this paid support activates the 12 reward center of the content creator's brain and releases dopamine while the creator is generating 13 revenue for YouTube. 14 c. Google's algorithms are designed to maximize "watch time." 15 741. Google engineers algorithms to recommend videos to YouTube users. 16 742. YouTube began building its' algorithms in 2008. 796 Its goal was to maximize how 17 long users spent watching YouTube videos. 797 18 743. These algorithms select videos that populate the YouTube homepage, rank results 19 in user searches, and push videos for viewers to watch through the "Up Next" feature. 20 744. YouTube designed its algorithms to manipulate users and induce them to use 21 YouTube excessively. 22 23 24 25 795 YouTube Partner Program: How to Make Money on YouTube, https://www.youtube.com/intl/en_us/creators/how-things-work/video-monetization/. 26 796 Cristos Goodrow, On YouTube's Recommendation System, YouTube (Sept. 15, 2021), https://blog.youtube/inside-youtube/on-youtubes-recommendation-system/. 27 797 Ben Popken, As Algorithms Take Over, YouTube's Recommendations Highlight a Human 28 Problem, NBC (Apr. 19, 2018), https://www.nbcnews.com/tech/social-media/algorithms-take- over-youtube-s-recommendations-highlight- human-problem-n867596. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 215 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 222 of 278 1 745. A former YouTube engineer explained that when he designed YouTube's algorithm, 2 YouTube wanted to optimize for one key metric: "watch time." 798 The engineer elaborated that 3 "[i]ncreasing users' watch time is good for YouTube's business model" because it increases 4 advertising revenue.799 5 746. In 2012 the YouTube Head of Content Creator Communications, similarly 6 explained: "When we suggest videos, we focus on those that increase the amount of time that the 7 viewer will spend watching videos on YouTube, not only on the next view, but also successive 8 views thereafter."800 9 747. The current algorithm uses deep-learning neural networks, a type of software that 10 returns outputs based on data fed into it. 801 The VP of Engineering at YouTube, explained that it is 11 "constantly evolving, learning every day from over 80 billion pieces of information [Google] calls 12 signals."802 Those signals include "watch and search history . . . , channel subscriptions, clicks, 13 watchtime, survey responses, and sharing, likes, and dislikes." 803 They also include user 14 demographic information like age and gender. 804 15 748. Google's algorithm also "uses data from your Google Account activity to influence 16 your recommendations."805 17 798 18 William Turton, How YouTube's Algorithm Prioritizes Conspiracy Theories, Vice (Mar. 5, 2018), https://www.vice.com/en/article/d3w9ja/how-youtubes-algorithm-prioritizes-conspiracy- 19 theories. 799 Jesselyn Cook & Sebastian Murdock, YouTube Is a Pedophile's Paradise, Huffington Post 20 (Mar. 20, 2020), https://www.huffpost.com/entry/youtube-pedophile- paradise_n_5e5d79d1c5b6732f50e6b4db. 21 800 Eric Meyerson, YouTube Now: Why We Focus on Watch Time, YouTube (Aug. 10, 2012), https://blog.youtube/news-and-events/youtube-now-why-we-focus-on-watch-time/. 22 801 Alexis C. Madrigal, How YouTube's Algorithm Really Works, Atlantic (Nov. 8, 2018), 23 https://www.theatlantic.com/technology/archive/2018/11/how-youtubes-algorithm-really- works/575212/; Paul Covington et al., Deep Neural Networks for YouTube Recommendations, 24 Google (2016), https://storage.googleapis.com/pub-tools-public-publication-data/pdf/45530.pdf. 802 Cristos Goodrow, On YouTube's Recommendation System, YouTube (Sept. 15, 2021), 25 https://blog.youtube/inside-youtube/on-youtubes-recommendation-system/. 803 Cristos Goodrow, On YouTube's Recommendation System, YouTube (Sept. 15, 2021), 26 https://blog.youtube/inside-youtube/on-youtubes-recommendation-system/. 804 Paul Covington et al., Deep Neural Networks for YouTube Recommendations, Google (2016), 27 https://storage.googleapis.com/pub-tools-public-publication-data/pdf/45530.pdf. 805 28 Manage Your Recommendations and Search Results, Google, https://support.google.com/youtube/answer/6342839?hl=en&co=GENIE.Platform%3DAndroid. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 216 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 223 of 278 1 749. The algorithm "develops dynamically" to predict which posts will hold the user's 2 attention.806 That is, it can also determine which "signals" are more important to individual users. 3 For example, if a user shares every video they watch, including those they rate low, the algorithm 4 learns to discount the significance of the user's shares when recommending content. 807 5 750. Besides the algorithm's self-learning capability, Google also consistently refines the 6 algorithm, updating it "multiple times a month." 808 7 751. In 2017, the former technical lead for YouTube recommendations explained that 8 "one of the key things [the algorithm] does is it's able to generalize." 809 While older iterations 9 "were pretty good at saying, here's another [video] just like" ones the user had watched, by 2017, 10 the algorithm could discern "patterns that are less obvious," identifying "adjacent relationships" of 11 "similar but not exactly the same" content. 810 12 752. Over time, the algorithm became increasingly successful in getting users to watch 13 recommended content. By 2018, YouTube Chief Product Officer Neal Mohan said that the 14 YouTube algorithm was responsible for more than 70% of users' time using the product. 811 That 15 is, more than 70% of the time users spend on YouTube was from recommendations Google's 16 algorithm pushed to them rather than videos identified by users through independent searches. 17 18 19 20 806 Cristos Goodrow, On YouTube's Recommendation System, YouTube (Sept. 15, 2021), https://blog.youtube/inside-youtube/on-youtubes-recommendation-system/. 21 807 Cristos Goodrow, On YouTube's Recommendation System, YouTube (Sept. 15, 2021), 22 https://blog.youtube/inside-youtube/on-youtubes-recommendation-system/. 808 Nilay Patel, YouTube Chief Product Officer Neal Mohan on The Algorithm, Monetization, and 23 the Future for Creators, Verge (Aug. 3, 2021), https://www.theverge.com/22606296/youtube- shorts-fund-neal-mohan-decoder-interview. 24 809 Casey Newton, How YouTube Perfected the Feed, Verge (Aug. 30, 2017), https://www.theverge.com/2017/8/30/16222850/youtube-google-brain-algorithm-video- 25 recommendation-personalized-feed. 810 Casey Newton, How YouTube Perfected the Feed, Verge (Aug. 30, 2017), 26 https://www.theverge.com/2017/8/30/16222850/youtube-google-brain-algorithm-video- recommendation-personalized-feed. 27 811 Joan E. Solsman, YouTube's AI Is the Puppet Master over Most of What You Watch, CNET 28 (Jan. 20, 2018), https://www.cnet.com/tech/services-and-software/youtube-ces-2018-neal- mohan/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 217 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 224 of 278 1 753. The algorithm also keeps users watching for longer periods. For instance, Mohan 2 explained that mobile device users watch for more than 60 minutes on average per session "because 3 of what our recommendations engines are putting in front of [them]." 812 4 754. The algorithm is particularly effective at addicting teenagers to the product. In 2022, 5 Pew Research Center found that "[a]bout three-quarters of teens visit YouTube daily, including 6 19% who report using the site or app almost constantly." 813 7 755. A software engineer explained that the algorithm is "an addiction engine." 814 He 8 raised concerns with YouTube staff, who said they had no intention to change the algorithms. After 9 all, the engineer explained, the algorithm works as intended: "it makes a lot of money." 815 10 756. Since users watch more than one billion hours of YouTube videos daily and 11 approximately 70% of the time is spent on videos pushed to users by YouTube's "recommendation 12 engine," Google's algorithms are responsible for hundreds of millions of hours users spend 13 watching videos on YouTube each day. 816 14 757. The videos pushed out to users by Google's "recommendation engine" are more 15 likely to be addictive and more likely to lead to harm. For example, "fear-inducing videos cause 16 the brain to receive a small amount of dopamine," which acts as a reward and creates a desire to do 17 something over and over.817 That dopaminergic response makes it more likely that a user will watch 18 the harmful video, which the algorithm interprets as signaling interest and preference. Former 19 812 Joan E. Solsman, YouTube's AI Is the Puppet Master over Most of What You Watch, CNET 20 (Jan. 20, 2018), https://www.cnet.com/tech/services-and-software/youtube-ces-2018-neal-mohan/ 813 21 Emily Vogels et al., Teens, Social Media and Technology 2022, Pew Rsch. Ctr. (Aug. 10, 2022), https://www.pewresearch.org/internet/2022/08/10/teens-social-media-and-technology- 22 2022. 814 Mark Bergen, YouTube Executives Ignored Warnings, Letting Toxic Videos Run Rampant, 23 Bloomberg (Apr. 2, 2019), https://www.bloomberg.com/news/features/2019-04-02/youtube- executives-ignored-warnings-letting-toxic-videos-run-rampant. 24 815 Mark Bergen, YouTube Executives Ignored Warnings, Letting Toxic Videos Run Rampant, Bloomberg (Apr. 2, 2019), https://www.bloomberg.com/news/features/2019-04-02/youtube- 25 executives-ignored-warnings-letting-toxic-videos-run-rampant. 816 See Joan E. Solsman, YouTube's AI Is the Puppet Master over Most of What You Watch, 26 CNET (Jan. 10, 2018), https://www.cnet.com/tech/services-and-software/youtube-ces-2018-neal- mohan/. 27 817 Josephine Bila, YouTube's Dark Side Could be Affecting Your Child's Mental Health, CNBC 28 (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization- in-young-children.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 218 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 225 of 278 1 Google engineers told the Wall Street Journal that "[t]he algorithm doesn't seek out extreme 2 videos . . . but looks for clips that data show are already drawing high traffic and keeping people 3 on the site. Those videos often tend to be sensationalist."818 An investigation by Bloomberg put it 4 simply: "In the race to one billion hours, a formula emerged: Outrage equals attention." 819 5 758. Google's algorithm makes it more likely for children to encounter harmful content 6 by pushing them down "rabbit holes," which "[lead] viewers to incrementally more extreme videos 7 or topics, which . . . hook them in."820 For example, a user might "[w]atch clips about bicycling, 8 and YouTube might suggest shocking bike race crashes." 821 In this way, the algorithm makes it 9 more likely that youth will encounter content that is violent, sexual, or encourages self-harm, 10 among other types of harmful content. 11 759. YouTube's "recommendation engine" creates a vicious cycle in its ruthless quest to 12 grow view time. Users who get pushed down rabbit holes then become models for the algorithm. 13 And the algorithm consequently emphasizes that harmful content, disproportionately pushing it to 14 more users. That is, because Google designed the algorithm to "maximize engagement," 15 uncommonly engaged users become "models to be reproduced." 822 Thus, the algorithms will "favor 16 the content of such users," which is often more extreme. 823 17 760. The algorithm also makes extreme content less likely to get flagged or reported. As 18 Guillaume Chaslot explained, the algorithm becomes "more efficient" over time "at recommending 19 818 Why is YouTube Suggesting Extreme and Misleading Content (2/7/2018), https://www.youtube.com/watch?v=7AjA3Df6i6o; see also Josephine Bila, YouTube's Dark Side 20 Could be Affecting Your Child's Mental Health, CNBC (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization-in-young- 21 children.html. 819 22 Mark Bergen, YouTube Executives Ignored Warnings, Letting Toxic Videos Run Rampant, Bloomberg (Apr. 2, 2019), https://www.bloomberg.com/news/features/2019-04-02/youtube- 23 executives-ignored-warnings-letting-toxic-videos-run-rampant. 820 Max Fisher & Amanda Taub, On YouTube's Digital Playground, an Open Gate for 24 Pedophiles, NY Times (June 3, 2019), https://www.nytimes.com/2019/06/03/world/americas/youtube-pedophiles.html. 25 821 Max Fisher & Amanda Taub, On YouTube's Digital Playground, an Open Gate for Pedophiles, NY Times (June 3, 2019), 26 https://www.nytimes.com/2019/06/03/world/americas/youtube-pedophiles.html. 822 Guillaume Chaslot, The Toxic Potential of YouTube's Feedback Loop, Wired (Jul. 13, 2019), 27 https://www.wired.com/story/the-toxic-potential-of-youtubes-feedback-loop/. 823 28 Guillaume Chaslot, The Toxic Potential of YouTube's Feedback Loop, Wired (Jul. 13, 2019), https://www.wired.com/story/the-toxic-potential-of-youtubes-feedback-loop/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 219 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 226 of 278 1 specific user-targeted content."824 And as the algorithm improves, "it will be able to more precisely 2 predict who is interested in [harmful or extreme] content." 825 So "problems with the algorithm 3 become exponentially harder to notice, as [harmful] content is unlikely to be flagged or 4 reported."826 5 761. Even on YouTube Kids, Google's product designed for children under 13 years old, 6 researchers from the Tech Transparency Project found that the product's algorithm fed children 7 content related to drugs and guns, as well as beauty and diet tips that risked creating harmful body 8 image issues. For example, the researchers found videos speaking positively about cocaine and 9 crystal meth; instructing users, step-by-step, how to conceal a gun; explaining how to bleach one's 10 face at home; and stressing the importance of burning calories. 827 11 12 13 14 15 16 17 18 19 20 21 22 23 824 24 Guillaume Chaslot, The Toxic Potential of YouTube's Feedback Loop, Wired (Jul. 13, 2019), https://www.wired.com/story/the-toxic-potential-of-youtubes-feedback-loop/. 25 825 Guillaume Chaslot, The Toxic Potential of YouTube's Feedback Loop, Wired (Jul. 13, 2019), https://www.wired.com/story/the-toxic-potential-of-youtubes-feedback-loop/. 26 826 Guillaume Chaslot, The Toxic Potential of YouTube's Feedback Loop, Wired (Jul. 13, 2019), https://www.wired.com/story/the-toxic-potential-of-youtubes-feedback-loop/. 27 827 Guns, Drugs, and Skin Bleaching: YouTube Kids Poses Risks to Children, Tech Transparency 28 Project (May 5, 2022), https://www.techtransparencyproject.org/articles/guns-drugs-and-skin- bleaching-youtube-kids-still-poses-risks-children. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 220 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 227 of 278 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 762. Amy Kloer, a campaign director with the child safety group Parents Together, spent 16 an hour on her preschool-age child's YouTube Kids account and found videos "encouraging kids 17 how to make their shirts sexier, a video in which a little boy pranks a girl over her weight, and a 18 video in which an animated dog pulls objects out of an unconscious animated hippo's butt." 828 19 Another parent recounted how YouTube Kids autoplay feature led her 6-year-old daughter to "an 20 animated video that encouraged suicide."829 21 763. These are not isolated examples. According to Pew Research Center, 46% of parents 22 of children 11 or younger report that children encountered videos that were inappropriate for their 23 age.830 And kids do not "choose" to encounter those inappropriate videos—YouTube's algorithm— 24 25 828 Rebecca Heilweil, YouTube's Kids App Has a Rabbit Hole Problem, Vox (May 12, 2021), https://www.vox.com/recode/22412232/youtube-kids-autoplay 26 829 Rebecca Heilweil, YouTube's Kids App Has a Rabbit Hole Problem, Vox (May 12, 2021), 27 https://www.vox.com/recode/22412232/youtube-kids-autoplay 830 Brooke Auxier et al., Parenting Children in The Age of Screens, Pew Rsch. Ctr. (July 28, 28 2020), https://www.pewresearch.org/internet/2020/07/28/parental-views-about-youtube/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 221 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 228 of 278 1 its "recommendation engine"—directs and pushes them there. Again, YouTube's algorithm is 2 responsible for 70% of the time users spend using the product. 831 3 764. Other reports have confirmed that YouTube's algorithm pushes users towards 4 harmful conduct. In 2021, the Mozilla Foundation studied 37,000 YouTube users, finding that 71% 5 of all reported negative user experiences came from videos recommended to users by Google's 6 algorithm. 832 And users were 40% more likely to report a negative experience from a video 7 recommended by YouTube's algorithm than from one they searched for. 833 Importantly, videos 8 that elicited those negative experiences "acquired 70% more views per day than other videos 9 watched by [study] volunteers."834 10 765. Those defects combine to compel children and teenagers to overuse a product that 11 feeds them harmful content, which in turn can adversely affect mental health. One 10-year-old girl 12 in the Mozilla Foundation study who sought "dance videos, ended up encountering videos 13 promoting extreme dieting."835 Her mother explained that "[s]he is now restricting her eating and 14 drinking."836 Another middle-schooler compulsively consumed YouTube videos every day after 15 she came home from school.837 Eventually, she became depressed and "got the idea to overdose 16 831 Joan E. Solsman, YouTube's AI Is the Puppet Master over Most of What You Watch, CNET 17 (Jan. 20, 2018), https://www.cnet.com/tech/services-and-software/youtube-ces-2018-neal- mohan/. 18 832 YouTube Regrets: A Crowdsourced Investigations into YouTube's Recommendation 19 Algorithm, Mozilla Foundation 13 (July 2021), https://assets.mofoprod.net/network/documents/Mozilla_YouTube_Regrets_Report.pdf. 20 833 YouTube Regrets: A Crowdsourced Investigations into YouTube's Recommendation Algorithm, Mozilla Foundation at 3 (July 2021), 21 https://assets.mofoprod.net/network/documents/Mozilla_YouTube_Regrets_Report.pdf. 834 22 YouTube Regrets: A Crowdsourced Investigations into YouTube's Recommendation Algorithm, Mozilla Foundation at 3 (July 2021), 23 https://assets.mofoprod.net/network/documents/Mozilla_YouTube_Regrets_Report.pdf. 835 YouTube Regrets: A Crowdsourced Investigations into YouTube's Recommendation 24 Algorithm, Mozilla Foundation at 13 (July 2021), https://assets.mofoprod.net/network/documents/Mozilla_YouTube_Regrets_Report.pdf. 25 836 YouTube Regrets: A Crowdsourced Investigations into YouTube's Recommendation 26 Algorithm, Mozilla Foundation at 13 (July 2021), https://assets.mofoprod.net/network/documents/Mozilla_YouTube_Regrets_Report.pdf. 27 837 Lesley McClurg, After Compulsively Watching YouTube, Teenage Girl Lands in Rehab for 'Digital Addiction', PBS (May 16, 2017), https://www.pbs.org/newshour/health/compulsively- 28 watching-youtube-teenage-girl-lands-rehab-digital-addiction. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 222 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 229 of 278 1 online."838 Three weeks later, she "down[ed] a bottle of Tylenol." She landed in rehab for digital 2 addiction due to her compulsive YouTube watching.839 3 766. Those experiences are not unique. Mental health experts have warned that YouTube 4 is a growing source of anxiety and inappropriate sexual behavior among kids under 13 years old. 5 Natasha Daniels, a child psychotherapist, described treating children between 8 and 10 years old, 6 who were "found doing sexual things: oral sex, kissing and getting naked and acting out sexual 7 poses."840 This kind of behavior "usually indicates some sort of sexual abuse." 841 Previously, 8 Daniels would typically "find a child who has been molested himself or that an adult has been 9 grooming the child from abuse." 842 But "in the last five years, when I follow the trail all the way 10 back, it's YouTube and that's where it ends." 843 11 767. Daniels has also seen increased rates of anxiety among children using YouTube. 12 And because of that anxiety, those children "exhibit loss of appetite, sleeplessness, crying fits, and 13 14 15 16 17 18 838 Lesley McClurg, After Compulsively Watching YouTube, Teenage Girl Lands in Rehab for 19 'Digital Addiction', PBS (May 16, 2017), https://www.pbs.org/newshour/health/compulsively- watching-youtube-teenage-girl-lands-rehab-digital-addiction. 20 839 Lesley McClurg, After Compulsively Watching YouTube, Teenage Girl Lands in Rehab for 'Digital Addiction', PBS (May 16, 2017), https://www.pbs.org/newshour/health/compulsively- 21 watching-youtube-teenage-girl-lands-rehab-digital-addiction. 840 22 Josephine Bila, YouTube's Dark Side Could be Affecting Your Child's Mental Health, CNBC (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization- 23 in-young-children.html. 841 Josephine Bila, YouTube's Dark Side Could be Affecting Your Child's Mental Health, CNBC 24 (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization- in-young-children.html. 25 842 Josephine Bila, YouTube's Dark Side Could be Affecting Your Child's Mental Health, CNBC 26 (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization- in-young-children.html. 27 843 Josephine Bila, YouTube's Dark Side Could be Affecting Your Child's Mental Health, CNBC (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization- 28 in-young-children.html. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 223 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 230 of 278 1 fear."844 Ultimately, she says, "YouTube is an ongoing conversation in my therapy practice, which 2 indicates there's a problem."845 3 768. One study determined that using Google's product was "consistently associated with 4 negative sleep outcomes."846 Specifically, for every 15 minutes teens spent using YouTube, they 5 were 24% less likely to get seven hours of sleep. According to Dr. Alon Avidan, director of the 6 UCLA Sleep Disorders Center, YouTube is particularly sleep disruptive because its 7 recommendation algorithm and autoplay features make it "so easy to finish one video" and watch 8 the next.847 Similarly, a signal that the YouTube algorithm relies on is the 'time of day' a user is 9 watching—a signal that, when used to maximize length of duration with the YouTube product, 10 induces sleep deprivation.848 11 769. Sleep deprivation is, in turn, associated with poor health outcomes. For example, 12 "insufficient sleep negatively affects cognitive performance, mood, immune function, 13 cardiovascular risk, weight, and metabolism."849 14 770. Compulsively consuming harmful content on YouTube can also harm brain 15 development. According to Donna Volpitta, Ed.D, "[c]hildren who repeatedly experience stressful 16 and/or fearful emotions may under develop parts of their brain's prefrontal cortex and frontal lobe, 17 18 844 19 Josephine Bila, YouTube's Dark Side Could be Affecting Your Child's Mental Health, CNBC (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization- 20 in-young-children.html. 845 Josephine Bila, YouTube's Dark Side Could be Affecting Your Child's Mental Health, CNBC 21 (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization- in-young-children.html. 22 846 Meg Pillion et al., What's 'app'-ning to adolescent sleep? Links between device, app use, and 23 sleep outcomes, 100 Sleep Med. 174–82 (Dec. 2022), https://www.sciencedirect.com/science/article/abs/pii/S1389945722010991?via%3Dihub. 24 847 Cara Murez, One App Is Especially Bad for Teens' Sleep, U.S. News & World Rep. (Sept. 13, 2022), https://www.usnews.com/news/health-news/articles/2022-09-13/one-app-is-especially- 25 bad-for-teens-sleep. 848 26 YouTube, How YouTube Works, https://www.youtube.com/howyoutubeworks/product- features/recommendations/#signals-used-to-recommend-content. 27 849 Jessica C. Levenson et al., The Association Between Social Media Use and Sleep Disturbance Among Young Adults, 85 Preventive Med. 36–41 (Apr. 2016), 28 https://www.sciencedirect.com/science/article/abs/pii/S0091743516000025. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 224 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 231 of 278 1 the parts of the brain responsible for executive functions, like making conscious choices and 2 planning ahead."850 3 771. Google's algorithm also promotes the creation of and pushes children towards 4 extremely dangerous prank or "challenge" videos, which often garner thousands of "Likes," adding 5 to the pressure chidren feel to participate. 851 For example, the YouTube algorithm repeatedly 6 pushed 10-year-old MDL plaintiff K.L.J. to videos of a viral prank called the "I Killed Myself 7 Prank," in which children pretend to have committed suicide to record their loved ones' reactions. 8 When K.L.J. eventually participated in the prank and tried to pretend to hang himself, he 9 accidentally did hang himself, suffering brain damage as a result. The neurological and 10 psychological techniques by which Google, like other Defendants, fosters excessive, addictive use 11 of YouTube in turn foster watching "challenge" videos. 12 772. Even though Google knew or should have known of these risks to its youth users, 13 Google's product lacks any warnings that foreseeable product use could cause these harms. 14 773. And despite all the evidence that YouTube's design and algorithms harm millions 15 of children, Google continues to manipulate users and compel them to use the product excessively, 16 to enhance Google's bottom line. As a result, young people are confronted with more and more 17 extreme videos, often resulting in significant harm. 18 d. YouTube's defective features include impediments to discontinuing use. 19 20 774. As with other Defendants, Google has intentionally and defectively designed its 21 products so that adolescent users, including Plaintiffs, face significant navigational obstacles and 22 hurdles when trying to delete or deactivate their accounts, in contrast to the ease with which users 23 can create those accounts. 24 25 850 26 Josephine Bila, YouTube's Dark Side Could be Affecting Your Child's Mental Health, CNBC (Feb. 13, 2018), https://www.cnbc.com/2018/02/13/youtube-is-causing-stress-and-sexualization- 27 in-young-children.html. 851 See, e.g., ViralBrothers, Revenge 9 – Cheating Prank Turns into Suicide Prank, YouTube 28 (June 11, 2014), https://www.youtube.com/watch?v=Bf7xIjz_ww0. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 225 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 232 of 278 1 775. First, because YouTube is accessible without a user needing to log in, YouTube 2 users cannot prevent themselves from being able to access YouTube by deleting their YouTube 3 account. 4 776. Second, YouTube accounts are linked to a user's broader Google account. These 5 accounts are structured such that, for a user to delete a YouTube account, the user must also delete 6 the user's entire Google account. This means that if a YouTube user uses Google's other products 7 those accounts will be lost as well. This structure holds hostage user data—if a child needs to keep 8 their email account through Google (for instance, if that is a requirement of their school), they 9 cannot delete their YouTube account, even if they want to. If a user stores family photos in Google 10 Photos, but wants to delete their YouTube account, they must choose between storage for their 11 photos or deleting their YouTube account. Similarly, if a user has purchased books or movies 12 through Google's digital market Google Play, the user's copy of those books or movies will be 13 deleted if the user deletes their Google account to rid themselves of YouTube. Google explicitly 14 threatens users with this consequence on the page where users can delete their account, listing every 15 associated account Google will delete and providing examples of the kinds of content that will be 16 deleted if a user does not back down from their desire to delete their YouTube account. 17 777. Third, Google intentionally designed its product so that to delete a user's Google 18 account, a user must locate and tap on six different buttons (through six different pages and popups) 19 from YouTube's main feed to delete an account successfully. This requires navigating away from 20 YouTube and into the webpages of other Google products. As with Meta, users are still able to 21 recover their accounts after deletion—though unlike Meta, Google does not tell users when their 22 accounts will become unrecoverable, simply threatening that they will soon after deletion. 23 4. Google facilitates the spread of CSAM and child exploitation. 24 778. Various design features of YouTube promote and dramatically exacerbate sexual 25 exploitation, the spread of CSAM, sextortion, and other socially maladaptive behavior that harms 26 children. 27 779. Google is required to comply with COPPA and obtain verifiable parental consent 28 before collecting personal information from children. It fails to do so. In 2019, the FTC and New PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 226 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 233 of 278 1 York Attorney General alleged in a federal complaint that Google and YouTube violated COPPA 2 by collecting personal information from children without verifiable parental consent. 852 3 780. Google and YouTube collected persistent identifiers that they used to track viewers 4 of child-directed channels across the Internet without prior parental notification, in violation of 5 Sections 1303(c), 1305(a)(1), and 1306(d) of COPPA. 853 6 781. Google and YouTube designed the child-centered YouTube Kids product. Despite 7 its clear knowledge of this channel being directed to children under 13 years old, Google served 8 targeted advertisements on these channels.854 9 782. Google pays its users to create content because it benefits from increased user 10 activity and receives something of value for its YouTube Partner Program. 855 11 783. Google allows users to monetize its product to generate revenue for itself and its 12 users, including users that violate laws prohibiting the sexual exploitation of children. 13 784. According to its own guidelines, Google prohibits using its social media product in 14 ways that "[endanger] the emotional and physical well-being of minors." 856 15 785. Google represents that YouTube "has strict policies and robust operations in place 16 to tackle content and behavior that is harmful or exploitative to children." 857 17 18 19 852 Fed. Trade Comm'n, Google and YouTube Will Pay Record $170 Million for Alleged Violations of Children's Privacy Law (2022), https://www.ftc.gov/news-events/news/press- 20 releases/2019/09/google-youtube-will-pay-record-170-million-alleged-violations-childrens- privacy-law. 21 853 Fed. Trade Comm'n, Google and YouTube Will Pay Record $170 Million for Alleged 22 Violations of Children's Privacy Law (2022), https://www.ftc.gov/news-events/news/press- releases/2019/09/google-youtube-will-pay-record-170-million-alleged-violations-childrens- 23 privacy-law. 854 Fed. Trade Comm'n, Google and YouTube Will Pay Record $170 Million for Alleged 24 Violations of Children's Privacy Law (2022), https://www.ftc.gov/news-events/news/press- releases/2019/09/google-youtube-will-pay-record-170-million-alleged-violations-childrens- 25 privacy-law. 855 YouTube Partner Program overview & eligibility, 26 https://support.google.com/youtube/answer/72851?hl=en. 856 Child safety policy - YouTube help, Google, 27 https://support.google.com/youtube/answer/2801999?hl=en. 857 28 Child safety policy - YouTube help, Google, https://support.google.com/youtube/answer/2801999?hl=en. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 227 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 234 of 278 1 786. Google maintains that its guidelines prohibit images, videos, and comments that put 2 children at risk, "including areas such as unwanted sexualization, abuse, and harmful and dangerous 3 acts."858 4 787. While Google "may place an age restriction on the video,"859 its product fails to 5 implement proper age-verification mechanisms to prevent minor users from accessing age- 6 restricted content, as discussed above. 7 788. Google fails to prevent collages of images and videos of children showing their 8 exposed buttocks, underwear, and genitals from racking up millions of views, on its product which 9 are then promoted and monetized by displaying advertisements from major brands alongside the 10 content.860 11 789. Through Google's product, videos of minors revealing their "bathing suit hauls," 12 playing in pools, beaches, waterparks, or performing gymnastics are recommended, shown, and 13 promoted to child predators who interact with these videos, including commenting to share "time 14 codes for crotch shots," to direct others to similar videos, and to arrange to meet up on other social 15 media products to share and exchange CSAM.861 16 790. Multiple YouTube channels dedicated to pre-teen models, young girls stretching, 17 and teen beauty are routinely oversexualized and manipulated by predators. 862 18 791. Google's product recommends and promotes abusive behaviors towards children 19 and victimizes unsuspecting minors on a mass scale. 20 792. When users search for images and videos of minors, Google's algorithm pushes 21 additional videos, which strictly feature children, and this recommended content often includes 22 promoted content for which Google receives value from advertisers. 23 858 Child safety policy - YouTube help, Google, 24 https://support.google.com/youtube/answer/2801999?hl=en. 859 Child safety policy - YouTube help, Google, 25 https://support.google.com/youtube/answer/2801999?hl=en. 860 K.G Orphanides, On YouTube, a network of pedophiles is hiding in plain sight WIRED UK 26 (2019), https://www.wired.co.uk/article/youtube-pedophile-videos-advertising. 861 K.G Orphanides, On YouTube, a network of pedophiles is hiding in plain sight WIRED UK 27 (2019), https://www.wired.co.uk/article/youtube-pedophile-videos-advertising. 862 28 K.G Orphanides, On YouTube, a network of pedophiles is hiding in plain sight WIRED UK (2019), https://www.wired.co.uk/article/youtube-pedophile-videos-advertising. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 228 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 235 of 278 1 793. Users of Google's product who search for images and videos of minors are further 2 inundated with comments from other predators that provide hyperlinks to CSAM and opportunities 3 to share CSAM on other products.863 4 794. On average, Google pays its creators $0.50 to $6.00 per 1,000 views of any video 5 they create, including materials depicting minors in violation of 18 U.S.C. §§ 2252, 2252A, 1591, 6 1466, and other criminal statutes.864 7 795. Google actively participates and receives value for creating content on its product 8 in violation of 18 U.S.C. §§ 2252, 2252A, 1591, 1466, and other criminal statutes. 9 796. Google actively participates and receives value for creating content on its product 10 in violation of laws prohibiting the sexual exploitation of children. 11 797. Google maintains that it is "dedicated to stopping the spread of online child 12 exploitation videos."865 Yet, it fails to implement proper safeguards to prevent the spread of illegal 13 contraband on its product. 14 798. The troves of data and information about its users that Google collects enable it to 15 detect, report as legally required, and take actions to prevent instances of sexual grooming, 16 sextortion, and CSAM distribution, but it has failed to do so. Google continues to make false 17 representations its "teams work around-the-clock to identify, remove, and report this content." 866 18 799. Google has proprietary technology, CSAI Match, that is supposed to combat CSAI 19 (Child Sexual Abuse Imagery) content online. This technology allows Google to identify known 20 CSAM contraband being promoted, shared, and downloaded on the YouTube product. Google's 21 CSAI Match can identify which portion of the video matches known and previously hashed CSAM 22 and provide a standardized categorization of the CSAM. When a match is detected by Google using 23 863 K.G Orphanides, On YouTube, a network of pedophiles is hiding in plain sight WIRED UK 24 (2019), https://www.wired.co.uk/article/youtube-pedophile-videos-advertising. 864 25 How YouTube creators earn money - how YouTube works, YouTube, https://www.youtube.com/howyoutubeworks/product- 26 features/monetization/#:~:text=Advertising%20is%20the%20primary%20way,directly%20profit %20from%20their%20work. 865 27 YouTube, https://www.youtube.com/csai-match/. 866 Google's efforts to combat online child sexual abuse material, 28 https://transparencyreport.google.com/child-sexual-abuse-material/reporting. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 229 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 236 of 278 1 CSAI Match, it is flagged so that Google can "responsibly report in accordance to local laws and 2 regulations."867 3 800. Despite this, Google routinely fails to flag CSAM and regularly fails to adequately 4 report known content to NCMEC and law enforcement, including CSAM depicting Plaintiffs, and 5 fails to takedown, remove, and demonetize CSAM. 6 801. Separate from CSAM detection, Google also implements an automated system 7 called Content ID "to easily identify and manage [its] copyright-protected content on YouTube." 868 8 Videos uploaded to YouTube are "scanned against a database of audio and visual content that's 9 been submitted to YouTube by copyright owners," and Google can block, monetize, and track that 10 material automatically.869 Google only grants Content ID to copyright owners who meet its own 11 specific criteria, and these criteria categorically exclude CSAM victims. Google fails to use Content 12 ID systems to block, remove, demonetize, or report CSAM on its product. 13 802. In 2018, Google launched "cutting-edge artificial intelligence (AI) that significantly 14 advances [Google's] existing technologies," which Google claimed "drastically improved" 15 detection of CSAM that is distributed by its YouTube product. 870 These claims were false, and 16 misled parents and children into believing its product is safe for minors. Google failed to drastically 17 improve the frequency of CSAM detection, reports, and takedowns on its product. 18 803. Google claims that it will "continue to invest in technology and organizations to 19 help fight the perpetrators of CSAM and to keep our products and our users safe from this type of 20 abhorrent content."871 In reality, it fails to do so. Google fails to invest in adequate age verification 21 and continues to fail to remove CSAM from its product. 22 867 Google's efforts to combat online child sexual abuse material, 23 https://transparencyreport.google.com/child-sexual-abuse-material/reporting. 868 How Content ID Works – YouTube Help, Google, 24 https://support.google.com/youtube/answer/2797370?hl=en. 869 How Content ID Works – YouTube Help, Google, 25 https://support.google.com/youtube/answer/2797370?hl=en. 870 Nikola Todorovic, Using AI to help organizations detect and report Child sexual abuse 26 material online Google (2018), https://blog.google/around-the-globe/google-europe/using-ai- help-organizations-detect-and-report-child-sexual-abuse-material-online/. 27 871 Nikola Todorovic, Using AI to help organizations detect and report Child sexual abuse 28 material online Google (2018), https://blog.google/around-the-globe/google-europe/using-ai- help-organizations-detect-and-report-child-sexual-abuse-material-online/. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 230 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 237 of 278 1 804. Google knows or should have known that YouTube facilitates the production, 2 possession, distribution, receipt, transportation, and dissemination of millions of materials that 3 depict obscene visual representations of the sexual abuse of children, or that violate child 4 pornography laws, each year. 5 805. Google knowingly fails to take adequate and readily available measures to remove 6 these contraband materials from its product in a timely fashion. 7 806. In violation of 18 U.S.C. § 2258A, Google knowingly fails to report massive 8 amounts of material in violation of 18 U.S.C. § 2256 and 18 U.S.C. § 1466A. 9 807. YouTube is polluted with illegal material that promotes and facilitates the sexual 10 exploitation of minors, and Google receives value in the form of increased user activity for the 11 dissemination of these materials on its products. 12 808. Google failed to report materials, including materials depicting Plaintiffs, in 13 violation of the reporting requirements of 18 U.S.C. § 2258A. 872 14 809. Google knows that its product is unsafe for children and yet fails to implement 15 safeguards to prevent children from accessing its product. 16 810. Further, there is effectively no way for users to report CSAM on Google's YouTube 17 product. YouTube does not allow users to specifically report any material posted on its product as 18 CSAM or child pornography.873 19 811. YouTube Mobile does not provide any way to report users, including users who 20 share CSAM on its product. On the desktop, a viewer can report a user, but Google has made the 21 reporting function difficult to access. Furthermore, reporting requires a viewer to have a Google 22 account and be logged in to the account to make the report. 874 23 24 872 NCMEC, 2019 CyberTipline reports by Electronic Service Providers (ESP), 25 https://www.missingkids.org/content/dam/missingkids/pdfs/2019-reports-by-esp.pdf. 873 Canadian Centre for Child Protection, Reviewing Child Sexual Abuse Material Reporting 26 Functions on Popular Platforms, https://protectchildren.ca/pdfs/C3P_ReviewingCSAMMaterialReporting_en.pdf. 27 874 Canadian Centre for Child Protection, Reviewing Child Sexual Abuse Material Reporting 28 Functions on Popular Platforms, at 18 https://protectchildren.ca/pdfs/C3P_ReviewingCSAMMaterialReporting_en.pdf. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 231 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 238 of 278 1 5. Google failed to adequately warn Plaintiffs about the harm its products cause or provide instructions regarding safe use. 2 3 812. Since YouTube's inception, Google has failed to adequately warn adolescent users 4 about the physical and mental health risks its product poses. These risks include, but are not limited 5 to, product abuse, addiction, and compulsive use; sexual exploitation from adult users; dissociative 6 behavior; damage to body image; social isolation; impaired brain development; and a plethora of 7 mental health disorders like body dysmorphia, eating disorders, anxiety, depression, insomnia, 8 ADD/ADHD exacerbation, suicidal ideation, self-harm, and death. 9 813. Google targets adolescent users via advertising and marketing materials distributed 10 throughout digital and traditional media products. Its advertising and marketing campaigns fail to 11 provide adequate warnings to potential adolescent consumers of the physical and mental risks 12 associated with using YouTube. 13 814. Google further fails to adequately warn adolescents during the product registration 14 process. At account setup, Google's product contains no warning labels, banners, or conspicuous 15 messaging to adequately inform adolescent users of the known risks and potential physical and 16 mental harms associated with usage of its product. Instead, Google allows adolescents to easily 17 create an account (or multiple accounts), and to access YouTube with or without an account. 18 815. Google's lack of adequate warnings continues once an adolescent uses YouTube. 19 Google does not adequately inform adolescent users that their data will be tracked, used to help 20 build a unique algorithmic profile, and potentially sold to Google's advertising clients. 21 816. Google's failure to warn adolescent users continues even as adolescents exhibit 22 problematic signs of addictive, compulsive use of YouTube. Google does not adequately warn users 23 when their screen time reaches harmful levels or when adolescents are accessing the product on a 24 habitual and uncontrolled basis. 25 817. Not only does Google fail to adequately warn users regarding the risks associated 26 with YouTube, it also does not provide adequate instructions on how adolescents can safely use its 27 product. A reasonable and responsible company would instruct adolescents on best practices and 28 safety protocols when using a product known to pose health risks. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 232 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 239 of 278 1 818. Google also fails to adequately warn users that: 2 sexual predators use YouTube to produce and distribute CSAM; 3 adult predators targeting young children for sexual exploitation, sextortion, 4 and CSAM are prevalent on YouTube; 5 usage of YouTube can increase the risk of children being targeted and 6 sexually exploited by adult predators; and, 7 usage of YouTube can increase risky and uninhibited behavior in children, 8 making them easier targets to adult predators for sexual exploitation, 9 sextortion, and CSAM. 10 819. Google failed to adequately warn parents about all of the foregoing dangers and 11 harms. Google's failure to adequately warn and instruct as set forth above has proximately caused 12 significant harm to Plaintiffs' and Consortium Plaintiffs' mental and physical well-being, and other 13 injuries and harms as set forth herein. 14 V. TIMELINESS AND TOLLING OF STATUTES OF LIMITATIONS 15 820. Through the exercise of reasonable diligence, Plaintiffs and Consortium Plaintiffs 16 did not and could not have discovered that Defendants' products caused their injuries and/or 17 sequelae thereto because, at the time of these injuries and/or sequelae thereto, the cause was 18 unknown to Plaintiffs and Consortium Plaintiffs. 19 821. Plaintiffs and Consortium Plaintiffs did not suspect and had no reason to suspect 20 Defendants' products caused his/her injuries and/or sequelae thereto until less than the applicable 21 limitations period prior to the filing of this action. 22 822. Due to the highly technical nature of the platforms' features, Plaintiffs and 23 Consortium Plaintiffs were unable to independently discovery that Defendants' products caused 24 their injuries and/or sequelae thereto until less than the applicable limitations period prior to the 25 filing of this action. 26 823. Defendants had exclusive knowledge of the material defects designed and 27 implemented into their platforms, and they have at all times through the present maintained their 28 proprietary designs of their platforms' features as strictly confidential. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 233 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 240 of 278 1 824. In addition, Defendants' fraudulent concealment and/or other tortious conduct has 2 tolled the running of any statute of limitations. 3 825. Defendants had a duty to disclose dangerous and defective features that cause 4 foreseeable harm to children and teens. 5 826. Defendants knowingly, affirmatively, and actively concealed from Plaintiffs and 6 Consortium Plaintiffs the risks associated with the defects of Defendants' products and that these 7 products caused their injuries and/or sequelae thereto. 8 827. Defendants committed tortious and/or fraudulent acts that continue to this day. As 9 of the date of this Complaint, Defendants still have not disclosed, and continue to conceal, that they 10 designed and implemented dangerous features into their platforms. Despite their knowledge of the 11 defects and their attendant safety risks, Defendants continue to market their platforms to children 12 and teens while simultaneously omitting the disclosure of known and foreseeable harms to children 13 and teens. 14 828. Plaintiffs were unaware and could not have reasonably known or learned through 15 reasonable diligence that they had been exposed to the defects and risks alleged herein and that 16 those defects and risks were the direct and proximate result of Defendants' acts and omissions. 17 829. Consortium Plaintiffs were unaware and could not have reasonably known or 18 learned through reasonable diligence that the harms they suffered were directly and proximately 19 caused by Defendants' acts and omissions. 20 830. For the foregoing reasons, Defendants are estopped from relying on any statutes of 21 limitation or repose as a defense in this action. All applicable statutes of limitation and repose have 22 been tolled by operation of the discovery rule and by Defendants' active concealment with respect 23 to all claims against Defendants. 24 VI. PLAINTIFFS' CLAIMS 25 831. Plaintiffs and Consortium Plaintiffs plead all Causes of Action of this Complaint in 26 the broadest sense, pursuant to all laws that may apply under choice-of-law principles, including 27 the law of the resident states of Plaintiffs and Consortium Plaintiffs. To the extent applicable to 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 234 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 241 of 278 1 specific Causes of Action, Plaintiffs and Consortium Plaintiffs plead these Causes of Action under 2 all applicable product liability acts, statutes, and laws of their respective States. 3 COUNT 1: STRICT LIABILITY – DESIGN DEFECT 4 (Against All Defendants) 5 832. Plaintiffs and Consortium Plaintiffs reallege and incorporate by reference each 6 preceding and succeeding paragraph as though set forth fully at length herein. 7 833. At all relevant times, each Defendant designed, developed, managed, operated, 8 tested, produced, labeled, marketed, advertised, promoted, controlled, sold, supplied, distributed, 9 and benefitted from its products used by Plaintiffs. 10 834. Each of the Defendant's respective products are designed and intended to be social 11 media products. 12 835. Each of the Defendant's respective products are distributed and sold to the public 13 through retail channels (i.e., the Apple App "Store" and the Google Play "Store"). 14 836. Each of the Defendant's respective products are marketed and advertised to the 15 public for the personal use of the end-user / consumer. 16 837. Each of the Defendant's defectively designed its respective products to addict 17 minors and young adults, who were particularly unable to appreciate the risks posed by the 18 products, and particularly susceptible to harms from those products. 19 838. The defects in the design of each of the Defendant's respective products existed 20 prior to the release of these products to Plaintiffs and the public, and there was no substantial change 21 to any of the Defendants' products between the time of their upload by each Defendant to public 22 or retail channels (e.g., the App Store or Google Play) and the time of their distribution to Plaintiffs 23 via download or URL access. 24 839. Plaintiffs used these products as intended, and each Defendant knew or, by the 25 exercise of reasonable care, should have known that Plaintiffs would use these products without 26 inspection for its addictive nature. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 235 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 242 of 278 1 840. Each Defendant defectively designed its respective products to take advantage of 2 the chemical reward system of users' brains (especially young users) to create addictive 3 engagement, compulsive use, and additional mental and physical harms. 4 841. Each Defendant failed to test the safety of the features it developed and implemented 5 for use on its respective products. When each Defendant did perform some product testing and had 6 knowledge of ongoing harm to Plaintiffs, it failed to adequately remedy its respective product's 7 defects or warn Plaintiffs. 8 842. Each of the Defendant's respective products are defective in design and pose a 9 substantial likelihood of harm for the reasons set forth herein, because the products fail to meet the 10 safety expectations of ordinary consumers when used in an intended or reasonably foreseeable 11 manner, and because the products are less safe than an ordinary consumer would expect when used 12 in such a manner. Children and teenagers are among the ordinary consumers of each of the 13 Defendant's products. Indeed, each Defendant markets, promotes, and advertises its respective 14 products to pre-teen and young consumers. Pre-teen and young consumers, and their parents and 15 guardians, do not expect Defendants' products to be psychologically and neurologically addictive 16 when the products are used in its intended manner by its intended audience. They do not expect the 17 algorithms and other features embedded by each Defendant in its respective products to make them 18 initially and progressively more stimulative, to maximize young consumers' usage time. They do 19 not expect each Defendant's revenues and profits to be directly tied to the strength of this addictive 20 mechanism and dependent on young consumers spending several hours a day using their respective 21 products. 22 843. Each of the Defendant's respective products are likewise defectively designed in 23 that it creates an inherent risk of danger; specifically, a risk of abuse, addiction, and compulsive 24 use by youth which can lead to a cascade of harms. Those harms include but are not limited to 25 dissociative behavior, withdrawal symptoms, social isolation, damage to body image and self- 26 worth, increased risky behavior, exposure to predators, sexual exploitation, and profound mental 27 health issues for young consumers including but not limited to depression, body dysmorphia, 28 anxiety, suicidal ideation, self-harm, insomnia, eating disorders, death, and other harmful effects. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 236 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 243 of 278 1 844. The risks inherent in the design of each of the Defendant's respective products 2 significantly outweigh any benefit of such design. 3 845. Each of the Defendants could have utilized cost-effective, reasonably feasible 4 alternative designs including algorithmic changes and changes to the addictive features described 5 above, to minimize the harms described herein, including, but not limited to: 6 Robust age verification; 7 Effective parental controls; 8 Effective parental notifications; 9 Warning of health effects of use and extended use upon sign-up; 10 Default protective limits to the length and frequency of sessions; 11 Opt-in restrictions to the length and frequency of sessions; 12 Self-limiting tools, including but not limited to session time notifications, 13 warnings, or reports; 14 Blocks to use during certain times of day (such as during school hours or late 15 at night); 16 Beginning and end to a user's "Feed;" 17 Redesigning algorithms to limit rather than promote addictive engagement; 18 Implementing labels on images and videos that have been edited through 19 product features such as "filters;" 20 Limits on the strategic timing and clustering of notifications to lure back 21 users; 22 Removing barriers to the deactivation and deletion of accounts; 23 Designing products that did not include the defective features listed in this 24 Complaint while still fulfilling the social networking purposes of a social 25 media product; 26 Implementing freely available and industry-proven child protection API 27 tools such as Project Arachnid Shield to help limit and prevent child sexual 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 237 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 244 of 278 1 exploitation, sextortion, and distribution of known CSAM through their 2 products; 3 Implementing reporting protocols to allow users or visitors of Defendants' 4 products to report CSAM and adult predator accounts specifically without 5 the need to create or log in to the products prior to reporting; 6 Implementing the legal definition of CSAM under 18 U.S.C. § 2256(8) and 7 related case law when scanning for CSAM using tools such as PhotoDNA 8 and CSAI to prevent underreporting of known CSAM; 9 Prioritizing "tolerance" rather than "efficiency" and "distinctness" of the 10 detection model when using scanning tools such as PhotoDNA and CSAI to 11 prevent underreporting of known CSAM; 12 Implementing client-side scanning and hashing and/or secure enclaves in the 13 direct messaging features of Meta's, Snap's, and ByteDance's products, to 14 prevent underreporting of known CSAM, and implementing proactive 15 detection measures to scan for known CSAM within all Defendants' social 16 media products and remove it; 17 Limiting or eliminating the use of geolocating for minors; 18 Eliminating product features that recommend minor accounts to adult 19 strangers; and 20 Others as set forth herein. 21 846. Alternative designs were available that would reduce minors' addictive and 22 compulsive engagement with each of the Defendants' respective products, and which would have 23 served effectively the same purpose of Defendants' products while reducing the gravity and 24 severity of danger posed by those products' defects. 25 847. Plaintiffs used Defendants' respective products as intended or in reasonably 26 foreseeable ways. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 238 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 245 of 278 1 848. The physical, emotional, and economic injuries of Plaintiffs and Consortium 2 Plaintiffs were reasonably foreseeable to each of the Defendants at the time of their respective 3 products' development, design, advertising, marketing, promotion, and distribution. 4 849. Defendants' respective products were defective and unreasonably dangerous when 5 they left the Defendants' respective possession and control. The defects continued to exist through 6 the products' distribution to and use by consumers, including Plaintiffs, who used the products 7 without any substantial change in the products' condition. 8 850. Plaintiffs and Consortium Plaintiffs were injured as a direct and proximate result of 9 each of the Defendant's respective defective designs as described herein. The defective design of 10 the products used by Plaintiffs was a substantial factor in causing harms to Plaintiffs and 11 Consortium Plaintiffs. 12 851. As a direct and proximate result of Defendants' respective products' defective 13 design, Plaintiffs and Consortium Plaintiffs suffered serious and dangerous injuries. 14 852. As a direct and proximate result of Defendants' respective products' defective 15 design, Plaintiffs and Consortium Plaintiffs require and/or will require more healthcare and services 16 and did incur medical, health, incidental, and related expenses. 17 853. The injuries of Plaintiffs and Consortium Plaintiffs cannot be wholly remedied by 18 monetary relief and such remedies at law are inadequate. 19 854. The nature of the fraudulent and unlawful acts that created safety concerns for 20 Plaintiffs are not the type of risks that are immediately apparent from using Defendants' respective 21 products. Many Plaintiffs are continuing to use Defendants' respective products. When Plaintiffs 22 use Defendants' respective products, they will not be independently able to verify whether 23 Defendants' respective products continue to pose an unreasonable risk or rely on Defendants' 24 respective representations in the future. 25 855. The conduct of each Defendant, as described above, was intentional, fraudulent, 26 willful, wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and 27 displayed an entire want of care and a conscious and depraved indifference to the consequences of 28 its conduct, including to the health, safety, and welfare of its customers, and warrants an award of PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 239 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 246 of 278 1 punitive damages in an amount sufficient to punish each Defendant and deter others from like 2 conduct. 3 856. Plaintiffs demand judgment against each Defendant for injunctive relief and for 4 compensatory, treble, and punitive damages, together with interest, costs of suit, attorneys' fees, 5 and all such other relief as the Court deems proper. 6 COUNT 2: STRICT LIABILITY – FAILURE TO WARN 7 (Against All Defendants) 8 857. Plaintiffs and Consortium Plaintiffs reallege and incorporate by reference each 9 preceding and succeeding paragraph as though set forth fully at length herein. 10 858. At all relevant times, each of the Defendants designed, developed, managed, 11 operated, tested, produced, labeled, marketed, advertised, promoted, controlled, sold, supplied, 12 distributed, and benefitted from its respective products used by Plaintiffs. 13 859. Plaintiffs were foreseeable users of Defendants' respective products. 14 860. Defendants' respective products are distributed and sold to the public through retail 15 channels (e.g., the Apple App "Store" and the Google Play "Store"). 16 861. Each of the Defendants sold and distributed its respective products to Plaintiffs in a 17 defective and unreasonably dangerous condition by failing to adequately warn about the risk of 18 harm to youth as described herein, including a risk of abuse, addiction, and compulsive use by 19 youth which can lead to a cascade of harms. Those harms include but are not limited to dissociative 20 behavior, withdrawal symptoms, social isolation, damage to body image and self-worth, increased 21 risky behavior, exposure to predators, sexual exploitation, and profound mental health issues for 22 young consumers including but not limited to depression, body dysmorphia, anxiety, suicidal 23 ideation, self-harm, insomnia, eating disorders, death, and other harmful effects. 24 862. Each of the Defendant's respective products is dangerous, to an extent beyond that 25 contemplated by the ordinary user who used Defendants' products, because they encourage 26 unhealthy, addictive engagement and compulsive use. 27 863. Each Defendant knew or, by the exercise of reasonable care, should have known 28 that its respective products posed risks of harm to youth considering its own internal data and PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 240 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 247 of 278 1 knowledge regarding its products at the time of development, design, marketing, promotion, 2 advertising, and distribution. 3 864. Defendants' respective products are defective and unreasonably dangerous because, 4 among other reasons described herein, each Defendant failed to exercise reasonable care to inform 5 users that, among other things: 6 Defendants' respective products cause addiction, compulsive use, and/or 7 other concomitant physical and mental injuries; 8 Defendants' respective products harvest and utilize user data in such a way 9 that increases a user's risk of addiction to these products and concomitant 10 physical and mental injuries; 11 The algorithmically-targeted feeds in Defendants' respective products are 12 designed to promote increasingly stimulative and alarming content to 13 encourage compulsive engagement by the user, raising the risk of mental 14 health harms including but not limited to depression, self-harm, and eating 15 disorders; 16 Defendants' respective products include features such as appearance- 17 altering "filters" that are known to promote negative social comparison, 18 body dysmorphia, and related injuries among youth by promoting artificial 19 and unrealistic beauty standards; 20 New users of Defendants' respective products can identify themselves as 21 minors, begin to use the product, and do so indefinitely, without any time or 22 usage limitations, without ever receiving a safety warning, and without ever 23 having to provide information so that each Defendant can warn the users' 24 parents or guardians; 25 The likelihood and severity of harms is greater for minors and young adults; 26 The likelihood and intensity of these harmful effects is exacerbated by the 27 interaction of each product's features with one another, and by algorithms 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 241 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 248 of 278 1 and other source code design that is currently publicly unknown and hidden 2 from the users and the government; 3 Sexual predators use Defendants' respective products to produce and 4 distribute CSAM; 5 Adult predators target young children for sexual exploitation, sextortion, and 6 CSAM on Defendants' respective products, with alarming frequency; 7 Usage of Defendants' respective products can increase the risk that children 8 are targeted and sexually exploited by adult predators; 9 Usage of Defendants' respective products can increase risky and uninhibited 10 behavior in children, making them easier targets to adult predators for sexual 11 exploitation, sextortion, and CSAM; and 12 End-to-end encryption and/or the ephemeral nature of Direct Messaging on 13 the Meta, ByteDance, and Snap products prevent the reporting of CSAM. 14 865. Ordinary users would not have recognized the potential risks of Defendants' 15 respective products when used in a manner reasonably foreseeable to each of the Defendants. 16 866. Had Plaintiffs received proper or adequate warnings or instructions as to the risks 17 of using Defendants' respective products, Plaintiffs would have heeded the warnings and/or 18 instructions. 19 867. Each of the Defendant's failures to adequately warn Plaintiffs about the risks of its 20 defective products was a proximate cause and a substantial factor in the injuries sustained by 21 Plaintiffs and Consortium Plaintiffs. 22 868. The injuries of Plaintiffs and Consortium Plaintiffs cannot be wholly remedied by 23 monetary relief and such remedies at law are inadequate. 24 869. The nature of the fraudulent and unlawful acts that created safety concerns for 25 Plaintiffs are not the type of risks that are immediately apparent from using Defendants' respective 26 products. Many Plaintiffs are continuing to use Defendants' respective products. When Plaintiffs 27 use Defendants' respective products, they will not be independently able to verify whether 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 242 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 249 of 278 1 Defendants' respective products continue to pose an unreasonable risk or rely on Defendants' 2 respective representations in the future. 3 870. The conduct of each Defendant, as described above, was intentional, fraudulent, 4 willful, wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and 5 displayed an entire want of care and a conscious and depraved indifference to the consequences of 6 its conduct, including to the health, safety, and welfare of their customers, and warrants an award 7 of punitive damages in an amount sufficient to punish each Defendant and deter others from like 8 conduct. 9 871. Plaintiffs and Consortium Plaintiffs demand judgment against each Defendant for 10 injunctive relief and for compensatory, treble, and punitive damages, together with interest, costs 11 of suit, attorneys' fees, and all such other relief as the Court deems proper. 12 COUNT 3: NEGLIGENCE – DESIGN 13 (Against All Defendants) 14 872. Plaintiffs and Consortium Plaintiffs reallege and incorporate by reference each 15 preceding and succeeding paragraph as though set forth fully at length herein. 16 873. At all relevant times, each of the Defendants designed, developed, managed, 17 operated, tested, produced, labeled, marketed, advertised, promoted, controlled, sold, supplied, 18 distributed, and benefitted from its respective products used by Plaintiffs. 19 874. Defendants' respective products were designed and intended to be social media 20 products. 21 875. Each of the Defendants knew or, by the exercise of reasonable care, should have 22 known, that its respective products were dangerous, harmful, and injurious when used by youth in 23 a reasonably foreseeable manner. 24 876. Each Defendant knew or, by the exercise of reasonable care, should have known 25 that its respective products posed risks of harm to youth. These risks were known and knowable in 26 light of each of the Defendant's own internal data and knowledge regarding its products at the time 27 of the products' development, design, marketing, promotion, advertising, and distribution to 28 Plaintiffs. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 243 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 250 of 278 1 877. Each of the Defendants knew, or by the exercise of reasonable care, should have 2 known, that ordinary consumers such as Plaintiffs would not have realized the potential risks and 3 dangers of the Defendants' respective products. Those risks include abuse, addiction, and 4 compulsive use in youth which can lead to a cascade of negative effects including but not limited 5 to dissociative behavior, withdrawal symptoms, social isolation, damage to body image and self- 6 worth, increased risky behavior, exposure to predators, sexual exploitation, and profound mental 7 health issues including but not limited to depression, body dysmorphia, anxiety, suicidal ideation, 8 self-harm, insomnia, eating disorders, and death. 9 878. Each of the Defendants owed a duty to all reasonably foreseeable users to design a 10 safe product. 11 879. Each of the Defendants owed a heightened duty of care to minor users of its 12 respective products because children's brains are not fully developed, resulting in a diminished 13 capacity to make responsible decisions regarding the frequency and intensity of social media usage. 14 Children are also more neurologically vulnerable than adults to the addictive aspects of Defendants' 15 respective products, such as the peer approval that comes from amassing follows and likes. 16 880. Each of the Defendants owe a particularly heightened duty of care to pre-teen users 17 (those under the age of 13), whose personal information is accorded special attention under federal 18 law. See 15 U.S.C. § 6501 et seq. 19 881. Plaintiffs were foreseeable users of the Defendants' respective products. 20 882. Each Defendant knew that minors such as Plaintiffs would use its respective 21 products. 22 883. Each Defendant breached its respective duty in designing its products. 23 884. Each Defendant breached its respective duty by failing to use reasonable care in the 24 design of its products by negligently designing them with features and algorithms as described 25 above that specifically are addictive and harmful to youth, who are particularly unable to appreciate 26 the risks posed by the products. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 244 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 251 of 278 1 885. Each Defendant breached its respective duty by designing products that were less 2 safe to use than an ordinary consumer would expect when used in an intended and reasonably 3 foreseeable manner. 4 886. Each Defendant breached its respective duty by failing to use reasonable care in the 5 design of its products by negligently designing its products with features and algorithms as 6 described above that created or increased the risk of abuse and addiction in youth, which can lead 7 to a cascade of negative effects including but not limited to dissociative behavior, withdrawal 8 symptoms, social isolation, damage to body image and self-worth, increased risky behavior, 9 exposure to predators, sexual exploitation, and profound mental health issues including but not 10 limited to depression, body dysmorphia, anxiety, suicidal ideation, self-harm, insomnia, eating 11 disorders, death, and other harmful effects. 12 887. Each Defendant breached its respective duty by failing to use reasonable care to use 13 cost-effective, reasonably feasible alternative designs, including algorithmic changes and changes 14 to the addictive features described above, and other safety measures, to minimize the harms 15 described herein. Alternative designs that would reduce the addictive features of Defendants' 16 respective products were available, would have served effectively the same purpose as each of the 17 Defendants' defectively designed products, and would have reduced the gravity and severity of 18 danger Defendants' respective products posed minor Plaintiffs. 19 888. A reasonable company under the same or similar circumstances as each Defendant 20 would have designed a safer product. 21 889. At all relevant times, Plaintiffs used Defendants' respective products in the manner 22 in which they were intended by Defendants to be used. 23 890. As a direct and proximate result of each of the Defendants' breached duties, 24 Plaintiffs and Consortium Plaintiffs were harmed. Defendants' design of their respective products 25 was a substantial factor in causing the Plaintiffs' and Consortium Plaintiffs' harms and injuries. 26 891. The injuries of Plaintiffs and Consortium Plaintiffs cannot be wholly remedied by 27 monetary relief and such remedies at law are inadequate. 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 245 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 252 of 278 1 892. The nature of the fraudulent and unlawful acts that created safety concerns for 2 Plaintiffs are not the type of risks that are immediately apparent from using Defendants' respective 3 products. Many Plaintiffs are continuing to use Defendants' respective products. When Plaintiffs 4 use Defendants' respective products, they will not be independently able to verify whether 5 Defendants' respective products continue to pose an unreasonable risk or rely on Defendants' 6 respective representations in the future. 7 893. The conduct of each Defendant, as described above, was intentional, fraudulent, 8 willful, wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and 9 displayed an entire want of care and a conscious and depraved indifference to the consequences of 10 its conduct, including to the health, safety, and welfare of its customers, and warrants an award of 11 punitive damages in an amount sufficient to punish each Defendant and deter others from like 12 conduct. 13 894. Plaintiffs and Consortium Plaintiffs demand judgment against each Defendant for 14 injunctive relief and for compensatory, treble, and punitive damages, together with interest, costs 15 of suit, attorneys' fees, and all such other relief as the Court deems proper. 16 COUNT 4: NEGLIGENCE – FAILURE TO WARN 17 (Against All Defendants) 18 895. Plaintiffs and Consortium Plaintiffs reallege and incorporate by reference each 19 preceding and succeeding paragraph as though set forth fully at length herein. 20 896. At all relevant times, each of the Defendants designed, developed, managed, 21 operated, tested, produced, labeled, marketed, advertised, promoted, controlled, sold, supplied, 22 distributed, and benefitted from its respective products used by Plaintiffs. 23 897. Plaintiffs were foreseeable users of Defendants' respective products. 24 898. Each of the Defendants knew, or by the exercise of reasonable care, should have 25 known, that use of their products was dangerous, harmful, and injurious when used in a reasonably 26 foreseeable manner, particularly by youth. 27 899. Each of the Defendants knew, or by the exercise of reasonable care, should have 28 known, that ordinary consumers such as Plaintiffs would not have realized the potential risks and PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 246 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 253 of 278 1 dangers of the Defendants' products including a risk of abuse, addiction, and compulsive use by 2 youth which can lead to a cascade of negative effects including but not limited to dissociative 3 behavior, withdrawal symptoms, social isolation, damage to body image and self-worth, and 4 profound mental health issues including but not limited to depression, body dysmorphia, anxiety, 5 suicidal ideation, self-harm, insomnia, eating disorders, and death. 6 900. Had Plaintiffs received proper or adequate warnings about the risks of Defendants' 7 respective products, Plaintiffs would have heeded such warnings. 8 901. Each of the Defendants knew or, by the exercise of reasonable care, should have 9 known that its products posed risks of harm to youth. These risks were known and knowable in 10 light of each of the Defendant's own internal data and knowledge regarding its products at the time 11 of development, design, marketing, promotion, advertising and distribution to Plaintiffs. 12 902. Each of the Defendants owed a duty to all reasonably foreseeable users, including 13 but not limited to minor users and their parents, to provide adequate warnings about the risk of 14 using Defendants' respective products that were known to each of the Defendants, or that each of 15 the Defendants should have known through the exercise of reasonable care. 16 903. Each of the Defendants owed a heightened duty of care to minor users and their 17 parents to warn about its products' risks because adolescent brains are not fully developed, resulting 18 in a diminished capacity to make responsible decisions regarding the frequency and intensity of 19 social media usage. Children are also more neurologically vulnerable than adults to the addictive 20 aspects of Defendants' respective products, including but not limited to the "flow state" created by 21 an endless feed and the public social validation created by follows and likes. 22 904. Each of the Defendants owe a particularly heightened duty of care to users under 23 the age of 13, whose personal information is accorded special protections under federal law. See 15 24 U.S.C. § 6501 et seq. 25 905. Each Defendant breached its duty by failing to use reasonable care in providing 26 adequate warnings to Plaintiffs and Consortium Plaintiffs, as set forth above. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 247 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 254 of 278 1 906. A reasonable company under the same or similar circumstances as Defendants 2 would have used reasonable care to provide adequate warnings to consumers, including the parents 3 of minor users, as described herein. 4 907. At all relevant times, each Defendant could have provided adequate warnings to 5 prevent the harms and injuries described herein. 6 908. As a direct and proximate result of each Defendant's breach of its respective duty to 7 provide adequate warnings, Plaintiffs and Consortium Plaintiffs were harmed and sustained the 8 injuries set forth herein. Each of the Defendants' failure to provide adequate and sufficient warnings 9 was a substantial factor in causing the harms to Plaintiffs and Consortium Plaintiffs. 10 909. As a direct and proximate result of each of the Defendants' failure to warn, Plaintiffs 11 and Consortium Plaintiffs require and/or will require more healthcare and services and did incur 12 medical, health, incidental, and related expenses. 13 910. The injuries of Plaintiffs and Consortium Plaintiffs cannot be wholly remedied by 14 monetary relief and such remedies at law are inadequate. 15 911. The nature of the fraudulent and unlawful acts that created safety concerns for 16 Plaintiffs are not the type of risks that are immediately apparent from using Defendants' respective 17 products. Many Plaintiffs are continuing to use Defendants' respective products. When Plaintiffs 18 use Defendants' respective products, they will not be independently able to verify whether 19 Defendants' respective products continue to pose an unreasonable risk or rely on Defendants' 20 respective representations in the future. 21 912. The conduct of each Defendant, as described above, was intentional, fraudulent, 22 willful, wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and 23 displayed an entire want of care and a conscious and depraved indifference to the consequences of 24 its conduct, including to the health, safety, and welfare of their customers, and warrants an award 25 of punitive damages in an amount sufficient to punish each Defendant and deter others from like 26 conduct. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 248 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 255 of 278 1 913. Plaintiffs and Consortium Plaintiffs demand judgment against each Defendant for 2 injunctive relief and for compensatory, treble, and punitive damages, together with interest, costs 3 of suit, attorneys' fees, and all such other relief as the Court deems proper. 4 COUNT 5: NEGLIGENCE 5 (Against All Defendants) 6 914. Plaintiffs and Consortium Plaintiffs reallege and incorporate by reference each 7 preceding and succeeding paragraph as though set forth fully at length herein, other than Counts 1- 8 4 (each of which is pled on a product theory). Count 5 is pled in the alternative on a non-product 9 theory. 10 915. At all relevant times, each Defendant developed, set up, managed, maintained, 11 operated, marketed, advertised, promoted, supervised, controlled, and benefitted from its respective 12 platforms used by Plaintiffs. 13 916. Each Defendant owed Plaintiffs a duty to exercise reasonable care in the 14 development, setup, management, maintenance, operation, marketing, advertising, promotion, 15 supervision, and control of its respective platforms not to create an unreasonable risk of harm from 16 and in the use of its platforms (including an unreasonable risk of addiction, compulsive use, sleep 17 deprivation, anxiety, depression, or other physical or mental injuries); to protect Plaintiffs from 18 unreasonable risk of injury from and in the use of its platforms; and not to invite, encourage, or 19 facilitate youth, such as Plaintiffs, to foreseeably engage in dangerous or risky behavior through, 20 on, or as a reasonably foreseeable result of using its platforms. 21 917. Plaintiffs were foreseeable users of the Defendants' respective platform(s). 22 918. Each Defendant knew that minors such as Plaintiffs would use their respective 23 platform(s). 24 919. Each Defendant invited, solicited, encouraged, or reasonably should have foreseen 25 the fact, extent, and manner of Plaintiffs' use of Defendants' respective platform(s). 26 920. Each Defendant knew or, by the exercise of reasonable care, should have known, 27 that the reasonably foreseeable use of its respective platforms (as developed, set up, managed, 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 249 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 256 of 278 1 maintained, supervised, and operated by that Defendant) was dangerous, harmful, and injurious 2 when used by youth such as Plaintiffs in a reasonably foreseeable manner. 3 921. At all relevant times, each Defendant knew or, by the exercise of reasonable care, 4 should have known that its respective platforms (as developed, setup, managed, maintained, 5 supervised, and operated by that Defendant) posed unreasonable risks of harm to youth such as 6 Plaintiffs, which risks were known and knowable, including in light of the internal data and 7 knowledge each Defendant had regarding its platforms. 8 922. Each Defendant knew, or by the exercise of reasonable care, should have known, 9 that ordinary youth users of its respective platforms, such as Plaintiffs, would not have realized the 10 potential risks and dangers of using the platform, including a risk of addiction, compulsive use, or 11 excessive use, which foreseeably can lead to a cascade of negative effects, including but not limited 12 to dissociative behavior, withdrawal symptoms, social isolation, damage to body image and self- 13 worth, increased risk behavior, exposure to predators, sexual exploitation and profound mental 14 health issues for young consumers including but not limited to depression, body dysmorphia, 15 anxiety, suicidal ideation, self-harm, insomnia, eating disorders, and death. 16 923. Each Defendant's conduct was closely connected to Plaintiffs' injuries, which were 17 highly certain to occur, as evidenced by the significance of Plaintiffs' injuries. 18 924. Each Defendant could have avoided Plaintiffs' injuries with minimal cost, 19 including, for example, by not including certain features and algorithms in its respective platforms 20 which harmed Plaintiffs. 21 925. Imposing a duty on Defendants would benefit the community at large. 22 926. Imposing a duty on Defendants would not be burdensome to them because they have 23 the technological and financial means to avoid the risks of harm to Plaintiffs. 24 927. Each Defendant owed a heightened duty of care to youth users of their respective 25 platforms because the child brain is not fully developed, meaning young people are more 26 neurologically vulnerable than adults to the addictive and other harmful aspects of Defendants' 27 respective platforms, and meaning young people have a diminished capacity to make responsible 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 250 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 257 of 278 1 decisions regarding the frequency, intensity, and manner of their use of Defendants' respective 2 platforms. 3 928. Each Defendant owes a particularly heightened duty of care to users under the age 4 of 13 due to the recognized safety risks posed to such users from interactive online platforms, such 5 as Defendants' respective products. See 15 U.S.C. § 6501 et seq. 6 929. Each Defendant has breached its duties of care owed to Plaintiffs through its 7 affirmative malfeasance, actions, business decisions, and policies in the development, setup, 8 management, maintenance, operation, marketing, advertising, promotion, supervision, and control 9 of its respective platforms. Those breaches include: 10 Including features and algorithms in their respective platforms that, as 11 described above, are currently structured and operated in a manner that 12 unreasonably creates or increases the foreseeable risk of addiction to, 13 compulsive use of, or overuse of the platform by youth, including Plaintiffs; 14 Including features and algorithms in their respective platforms that, as 15 described above, are currently structured and operated in a manner that 16 unreasonably creates or increases the foreseeable risk of harm to the physical 17 and mental health and well-being of youth users, including Plaintiffs, 18 including but not limited to dissociative behavior, withdrawal symptoms, 19 social isolation, depression, anxiety, suicide and suicidal ideation, body 20 dysmorphia, self-harm, sleep deprivation, insomnia, eating disorders, and 21 death; 22 Including features and algorithms in their respective platforms that, as 23 described above, are currently structured and operated in a manner that 24 unreasonably exposes youth users to sexual predators and sexual 25 exploitation, including features that recommend or encourage youth users to 26 connect with adult strangers on or through the platform; 27 Maintaining unreasonably dangerous features and algorithms in their 28 respective platforms after notice that such features and algorithms, as PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 251 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 258 of 278 1 structured and operated, posed a foreseeable risk of harm to the physical and 2 mental health and well-being of youth users; 3 Facilitating use of their respective platforms by youth under the age of 13, 4 including by adopting protocols that do not ask for or verify the age or 5 identity of users or by adopting ineffective age and identity verification 6 protocols; and 7 Facilitating unsupervised and/or hidden use of their respective platforms by 8 youth, including by adopting protocols that allow youth users to create 9 multiple and private accounts and by offering features that allow youth users 10 to delete, hide, or mask their usage. 11 930. Each Defendant has breached its duties of care owed to Plaintiffs through its non- 12 feasance, failure to act, and omissions in the development, setup, management, maintenance, 13 operation, marketing, advertising, promotion, supervision, and control of its respective platforms. 14 Those breaches include: 15 Failing to implement effective protocols to block users under the age of 13; 16 Failing to implement effective parental controls; 17 Failing to implement reasonably available means to monitor for and limit or 18 deter excessive frequency or duration of use of platforms by youth, including 19 patterns, frequency, or duration of use that are indicative of addiction, 20 compulsive use, or overuse; 21 Failing to implement reasonably available means to limit or deter use of 22 platforms by youth during ordinary times for school or sleep; 23 Failing to implement reasonably available means to set up and operate its 24 platforms without algorithms and features, discussed above, that rely on 25 unreasonably dangerous methods (such as endless scroll, autoplay, IVR, 26 social comparison, and others) as a means to engage youth users; 27 Failing to set up, monitor, and modify the algorithms used on their platforms 28 to prevent the platforms from actively driving youth users into unsafe, PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 252 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 259 of 278 1 distorted, and unhealthy online experiences, including highly sexualized, 2 violent, and predatory environments and environments promoting eating 3 disorders and suicide; 4 Failing to implement reasonably available means to monitor for, report, and 5 prevent the use of their platforms by sexual predators to victimize, abuse, 6 and exploit youth users; and 7 Failing to provide effective mechanisms for youth users and their 8 parents/guardians to report abuse or misuse of the platforms. 9 931. A reasonable company under the same or similar circumstances as each Defendant 10 would have developed, set up, managed, maintained, supervised, and operated its platforms in a 11 manner that is safer for and more protective of youth users like Plaintiffs. 12 932. At all relevant times, Plaintiffs used one or more of the Defendants' respective 13 platforms in the manner in which they were intended to be used. 14 933. As a direct and proximate result of each Defendant's breach of one or more of its 15 duties, Plaintiffs and Consortium Plaintiffs were harmed. Such harms include addiction to, or 16 compulsive or excessive use of, Defendants' platforms, and cascade of resulting negative effects, 17 including but not limited to dissociative behavior, withdrawal symptoms, social isolation, damage 18 to body image and self-worth, increased risky behavior, exposure to predators, sexual exploitation 19 and profound mental health issues including but not limited to depression, body dysmorphia, 20 anxiety, suicidal ideation, self-harm, insomnia, eating disorders, and death. 21 934. Each Defendant's breach of one or more of its duties was a substantial factor in 22 causing harms and injuries to the Plaintiffs and Consortium Plaintiffs. 23 935. The injuries of Plaintiffs and Consortium Plaintiffs cannot be wholly remedied by 24 monetary relief and such remedies at law are inadequate. 25 936. The nature of the fraudulent and unlawful acts that created safety concerns for 26 Plaintiffs are not the type of risks that are immediately apparent from using Defendants' respective 27 products. Many Plaintiffs are continuing to use Defendants' respective products. When Plaintiffs 28 use Defendants' respective products, they will not be independently able to verify whether PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 253 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 260 of 278 1 Defendants' respective products continue to pose an unreasonable risk or rely on Defendants' 2 respective representations in the future. 3 937. Each Defendant's conduct, as described above, was intentional, fraudulent, willful, 4 wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and displayed an 5 entire want of care and a conscious and depraved indifference to the consequences of their conduct, 6 including to the health, safety, and welfare of their customers, and warrants an award of punitive 7 damages in an amount sufficient to punish the Defendants and deter others from like conduct. 8 938. Plaintiffs and Consortium Plaintiffs demand judgment against each Defendant for 9 injunctive relief and for compensatory, treble, and punitive damages, together with interest, costs 10 of suit, attorneys' fees, and all such other relief as the Court deems proper. 11 COUNT 6: INTENTIONALLY OMITTED 12 13 939–955. Intentionally omitted. 14 COUNT 7: VIOLATION OF UNFAIR TRADE 15 PRACTICES/CONSUMER PROTECTION LAWS (Against All Defendants) 16 17 956. Plaintiffs and Consortium Plaintiffs reallege and incorporate by reference each 18 preceding and succeeding paragraph as though set forth fully at length herein. 19 957. At all relevant times, each of the Defendants designed, developed, managed, 20 operated, tested, produced, labeled, marketed, advertised, promoted, controlled, sold, supplied, 21 distributed, and benefitted from its respective products. 22 958. Certain Plaintiffs and/or Consortium Plaintiffs will bring a cause of action for 23 consumer fraud and/or unfair, deceptive, and abusive trade practices under applicable state law. 24 959. Each Defendant is on notice that such claims may be asserted by those Plaintiffs 25 and/or Consortium Plaintiffs. 26 960. Plaintiffs used Defendants' respective social media products and suffered 27 ascertainable losses and injuries as a result of each Defendant's respective actions in violation of 28 consumer protection laws. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 254 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 261 of 278 1 961. Had each Defendant not engaged in the respective unfair, deceptive, and abusive 2 conduct described herein, Plaintiffs would not have used Defendants' respective social media 3 products resulting in the monetary and physical injuries as alleged herein. 4 962. Fraudulent, unfair, deceptive, and/or abusive practices that violate state consumer 5 protection laws include but are not limited to the following: 6 representing that goods or services have approval, characteristics, uses, or 7 benefits that they do not have; 8 advertising goods or services with the intent not to sell them as advertised; 9 engaging in deception, fraud, false pretense, false premise, 10 misrepresentation, or knowing concealment, suppression, or omission of any 11 material fact with the intent that a consumer rely on the same in connection 12 with the promotion of products and/or that creates a likelihood of confusion; 13 and 14 Failing to provide adequate notice to parents about collecting personal 15 information of children under the age of 13, as required by COPPA and 16 accompanying regulations, the violation of which "shall be treated as a 17 violation of a rule defining an unfair or deceptive act or practice." 15 U.S.C. 18 §§ 6502(c). 19 963. Each Defendant has a statutory duty to refrain from fraudulent, unfair, deceptive, 20 and abusive acts or trade practices in the design, development, promotion, and distribution of its 21 respective products. 22 964. Each Defendant violated consumer protection laws, through its use of false and 23 misleading misrepresentations or omissions of material fact relating to the safety of its respective 24 social media products. 25 965. Each Defendant uniformly communicated the purported benefits and safety of using 26 its respective social media products while failing to disclose the serious harms related to use, 27 particularly to youth. Defendants made these representations to the public, the government, and 28 consumers. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 255 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 262 of 278 1 966. Under the various consumer protection laws in each state and other consumer 2 protection statutes, each Defendant is the supplier, distributor, advertiser, marketer, and promoter 3 of its respective social media products, and is subject to liability under such statutes for fraudulent, 4 unfair, deceptive, abusive, and unconscionable consumer sales practices. Defendants' actions and 5 omissions are uncured or incurable and Defendants were put on notice more than 30 days before 6 this filing and failed to take any action to cure their actions or omissions. 7 967. Defendants' deceptive, unconscionable, unfair, abusive, and/or fraudulent 8 representations and material omissions to Plaintiffs and Consortium Plaintiffs constituted consumer 9 fraud and/or unfair, deceptive, and abusive acts and trade practices in violation of the various 10 consumer protection statutes in the states and territories of the United States, which will be 11 identified in the Short Form Complaints. 12 968. Each Defendant had actual knowledge of the defective and dangerous condition of 13 its respective products and failed to take any action to cure those conditions. 14 969. The actions or omissions of each Defendant are uncured or incurable. 15 970. Plaintiffs and Consortium Plaintiffs relied to their detriment on Defendants' 16 respective misrepresentations and omissions in deciding to use Defendants' respective social media 17 products. 18 971. By reason of the fraudulent and unlawful acts engaged in by each of the Defendants, 19 and as a direct and proximate result thereof, Plaintiffs and Consortium Plaintiffs have sustained 20 economic losses and other damages and are entitled to statutory and compensatory damages in an 21 amount to be proven at trial. 22 972. The injuries of Plaintiffs and Consortium Plaintiffs cannot be wholly remedied by 23 monetary relief and such remedies at law are inadequate. 24 973. The nature of the fraudulent and unlawful acts that created safety concerns for 25 Plaintiffs are not the type of risks that are immediately apparent from using Defendants' respective 26 products. Many Plaintiffs are continuing to use Defendants' respective products. When Plaintiffs 27 use Defendants' respective products, they will not be independently able to verify whether 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 256 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 263 of 278 1 Defendants' respective products continue to pose an unreasonable risk or rely on Defendants' 2 respective representations in the future. 3 974. The conduct of each Defendant, as described above, was intentional, fraudulent, 4 willful, wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and 5 displayed an entire want of care and a conscious and depraved indifference to the consequences of 6 its conduct, including to the health, safety, and welfare of their customers, and warrants an award 7 of punitive damages in an amount sufficient to punish each Defendant and deter others from like 8 conduct. 9 975. Plaintiffs and Consortium Plaintiffs demand judgment against each of the 10 Defendants for injunctive and monetary relief, including compensatory, treble, and punitive 11 damages, together with interest, costs of suit, attorneys' fees, and all such other relief as the Court 12 deems proper. 13 COUNT 8: FRAUDULENT CONCEALMENT AND MISREPRESENTATION 14 (Against Meta) 15 976. Plaintiffs and Consortium Plaintiffs re-allege and incorporate by reference each 16 preceding and succeeding paragraph as though set forth fully at length herein. 17 977. This claim is brought against Meta. 18 978. As set forth in more detail above, Meta knew about the defective condition of 19 Instagram and Facebook and that the products posed serious health risks to users. 20 979. Meta was under a duty to tell the public the truth and to disclose the defective 21 condition of Instagram and Facebook and that the products posed serious health risks to users, 22 particularly youth. 23 980. Meta breached its duty to the public, users, and their parents, including Plaintiffs 24 and Consortium Plaintiffs, by concealing, failing to disclose, and making misstatements about the 25 serious safety risks presented by Instagram and Facebook. Even though Meta knew of those risks 26 based on Meta's internal studies, external studies known to Meta, and information conveyed by at 27 least one scientific expert directly to Mark Zuckerberg, it intentionally concealed those findings, in 28 order not to lose users and advertising revenue, and to induce youth, including Plaintiffs, to continue PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 257 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 264 of 278 1 using Instagram and Facebook. Worse still, Meta made numerous partial material representations 2 downplaying any potential harm associated with Instagram and Facebook and reassuring the public, 3 Congress, and parents that these products were safe. 4 981. By intentionally concealing and failing to disclose defects inherent in the design of 5 Instagram and Facebook, Meta knowingly and recklessly misled the public, users, and their parents, 6 including Plaintiffs and Consortium Plaintiffs, into believing these products were safe for children 7 to use. 8 982. By intentionally making numerous partial material representations, downplaying 9 any potential harm associated with Instagram and Facebook, and reassuring the public, Congress, 10 and parents, including Plaintiffs and Consortium Plaintiffs, that it was safe, Meta fraudulently 11 misled the public, users, and their parents, including Plaintiffs and Consortium Plaintiffs, into 12 believing Instagram and Facebook were safe for children to use. 13 983. Meta knew that its concealment, misstatements, and omissions were material. A 14 reasonable person, including Plaintiffs and Consortium Plaintiffs, would find information that 15 impacted the users' health, safety, and well-being, such as serious adverse health risks associated 16 with the use of Instagram and Facebook, to be important when deciding whether to use, or continue 17 to use, those products. 18 984. As a direct and proximate result of Meta's material omissions, misrepresentations, 19 and concealment of material information, Plaintiffs and Consortium Plaintiffs were not aware and 20 could not have been aware of the facts that Meta concealed or misstated, and therefore justifiably 21 and reasonably believed that Instagram and Facebook were safe for children to use. 22 985. As a direct and proximate result of Meta's material omissions, misrepresentations, 23 and concealment of material information, Plaintiffs and Consortium Plaintiffs sustained serious 24 injuries and harm. 25 986. Meta's conduct, as described above, was intentional, fraudulent, willful, wanton, 26 reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and displayed an entire want 27 of care and a conscious and depraved indifference to the consequences of its conduct, including to 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 258 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 265 of 278 1 the health, safety, and welfare of its customers, and warrants an award of punitive damages in an 2 amount sufficient to punish Meta and deter others from like conduct. 3 987. Plaintiffs and Consortium Plaintiffs demand judgment against Meta for 4 compensatory, treble, and punitive damages, together with interest, costs of suit, attorneys' fees, 5 and all such other relief as the Court deems proper. 6 COUNT 9: NEGLIGENT CONCEALMENT AND MISREPRESENTATION 7 (Against Meta) 8 988. Plaintiffs and Consortium Plaintiffs reallege and incorporate by reference each 9 preceding and succeeding paragraph as though set forth fully at length herein. 10 989. This claim is brought against Meta. 11 990. As set forth in more detail above, Meta knew about the defective condition of the 12 Instagram and Facebook products and that the products posed serious health risks to users, 13 particularly youth. 14 991. Meta was under a duty to tell the public the truth and to disclose the defective design 15 of Instagram and Facebook and that the products posed serious health risks to users. 16 992. Meta owed a heightened duty of care to minor users of its products because 17 children's brains are not fully developed, resulting in a diminished capacity to make responsible 18 decisions regarding the frequency and intensity of social media usage. Children are also more 19 neurologically vulnerable than adults to the addictive aspects of Instagram and Facebook, such as 20 the peer approval that comes from amassing follows and likes. 21 993. Meta breached its duty to the public, users, and their parents, including Plaintiffs 22 and Consortium Plaintiffs, and failed to take reasonable care by concealing, failing to disclose, and 23 making misstatements about the serious safety risks presented by its products. Even though Meta 24 knew of those risks based on Meta's internal studies, external studies known to Meta, and 25 information provided by at least one scientific expert directly to Zuckerberg, Meta negligently 26 concealed those findings, in order not to lose users and advertising revenue, and to induce children, 27 including Plaintiffs, to continue using its products. Worse still, Meta negligently made numerous 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 259 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 266 of 278 1 partial material representations downplaying any potential harm associated with its products and 2 reassuring the public and parents its products were safe. 3 994. By concealing and failing to disclose, or taking reasonable care to disclose the 4 defects, Meta negligently misled users and their parents, including Plaintiffs and Consortium 5 Plaintiffs, into believing Instagram and Facebook were safe for children to use. 6 995. By making numerous partial material representations downplaying any potential 7 harm associated with its products and reassuring the public, Congress, and parents, including 8 Plaintiffs and Consortium Plaintiffs, that its products were safe, Meta negligently misled the public 9 users and their parents, including Plaintiffs and Consortium Plaintiffs, into believing Meta's 10 products were safe for use. 11 996. As a direct and proximate result of Meta's material omissions, misrepresentations, 12 and concealment of material information, Plaintiffs and Consortium Plaintiffs were not aware and 13 could not have been aware of the facts that Meta concealed or misstated, and therefore justifiably 14 and reasonably believed that Instagram and Facebook were safe for use. 15 997. As a direct and proximate result of Meta's material omissions, misrepresentations, 16 and concealment of material information, Plaintiffs and Consortium Plaintiffs sustained serious 17 injuries and harm. 18 998. Meta's conduct, as described above, was intentional, fraudulent, willful, wanton, 19 reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and displayed an entire want 20 of care and a conscious and depraved indifference to the consequences of its conduct, including to 21 the health, safety, and welfare of their customers, and warrants an award of punitive damages in an 22 amount sufficient to punish them and deter others from like conduct. 23 999. Plaintiffs and Consortium Plaintiffs demand judgment against Meta for 24 compensatory, treble, and punitive damages, together with interest, costs of suit, attorneys' fees, 25 and all such other relief as the Court deems proper. 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 260 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 267 of 278 1 COUNT 10: NEGLIGENCE PER SE 2 (Against All Defendants) 3 1000. Plaintiffs and Consortium Plaintiffs re-allege and incorporate by reference each 4 preceding and succeeding paragraph as though set forth fully at length herein. 5 1001. At all times, each of the Defendants had an obligation to comply with applicable 6 statutes and regulations, including but not limited to the PROTECT Our Children Act (see 18 7 U.S.C. §§ 2258A, 2258B), and COPPA (see 15 U.S.C. §§ 6501-6506, 16 C.F.R. § 312.2), as well 8 as other federal laws. 9 1002. Each of the Defendants owed a heightened duty of care to minor users and their 10 parents to implement age verification services that were effective and would prevent access by pre- 11 teen users of its respective products. 12 1003. Each of the Defendant's respective actions as described herein violated these 13 statutes and regulations, as well as other federal laws. 14 1004. Each of the Defendants failed to meet the requirements of 18 U.S.C. § 2258A by not 15 reporting to NCMEC the violations of child pornography laws that they suspected to be in existence 16 within its respective products. 17 1005. Specifically, each of the Defendants intentionally designed its respective products 18 in an attempt to limit and avoid its reporting requirements under this statute. 19 1006. Each of the Defendants also failed to minimize the numbers of its respective 20 employees with access to visual depictions of Plaintiffs in violation of 18 U.S.C. § 2258B(c). 21 1007. COPPA (15 U.S.C. §§ 6501-6506; 16 C.F.R. § 312.2) establishes certain obligations 22 for operators of websites and online services that are intended to inform parents and guardians 23 about the collection of their children's personal information. 24 1008. Each Defendant is an "operator," as defined by 16 C.F.R. § 312.2. 25 1009. Each of the Defendants collects or uses personal information from children under 26 the age of 13 through its respective websites or online services that are directed to (or that each 27 Defendant has actual knowledge were used by) children. Each Defendant has actual knowledge 28 that it collects personal information directly from users of its respective websites or online services. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 261 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 268 of 278 1 1010. Each Defendant has collected or used personal information from children younger 2 than age 13 in violation of COPPA by, at least: 3 Failing to provide through their respective websites and apps a clear, 4 understandable, and complete direct notice to parents that described each 5 Defendant's respective practices regarding the collection, use, or disclosure 6 of children's personal information, in violation of 16 C.F.R. § 312.4(a) and 7 (c); 8 Failing to make reasonable efforts, taking into account available technology, 9 to ensure parents received such notice on their websites and applications so 10 that parents could provide informed consent, in violation of 16 C.F.R. § 11 312.4(b)-(c); and 12 Failing to obtain verifiable parental consent before any collection, use, or 13 disclosure of personal information from children, in violation of 16 C.F.R. 14 § 312.5(a)(1). 15 1011. Plaintiffs are within the class of persons that these statutes and regulations are 16 intended to protect. This includes Plaintiffs who, as minors who use the Internet, are within the 17 scope of persons the PROTECT Our Children Act and COPPA are intended to protect. 18 1012. Plaintiffs' injuries and/or symptoms are the type of harm that these statutes and 19 regulations are intended to prevent. 20 1013. Violations of the foregoing statutes and regulations, among others, by each 21 Defendant constitutes negligence per se. 22 1014. As a direct and proximate result of each of the Defendant's respective statutory and 23 regulatory violations, Plaintiffs and Consortium Plaintiffs suffered serious injuries and/or sequelae 24 thereto, including but not limited to emotional distress, diagnosed mental health conditions, loss of 25 income and earning capacity, reputational harm, physical harm, past and future medical expenses, 26 and pain and suffering. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 262 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 269 of 278 1 1015. As a direct and proximate result of each of the Defendants' respective statutory and 2 regulatory violations, Plaintiffs and Consortium Plaintiffs require and/or will require more 3 healthcare and services and did incur medical, health, incidental, and related expenses. 4 1016. Plaintiffs and Consortium Plaintiffs may also require additional medical and/or 5 hospital care, attention, and services in the future. 6 1017. As a result of each of the Defendant's negligence per se, Plaintiffs suffered severe 7 mental harm, leading to physical and mental injury, from use of and exposure to Defendants' 8 respective social media products. Plaintiffs and Consortium Plaintiffs suffered serious damages in 9 the form of emotional distress, diagnosed mental health conditions, medical expenses, loss of 10 income and earning capacity, pain and suffering, and reputational harm. 11 1018. Plaintiffs and Consortium Plaintiffs have suffered physical harm, emotional distress, 12 past and future medical expenses, and pain and suffering. 13 1019. The conduct of each Defendant, as described above, was intentional, fraudulent, 14 willful, wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and 15 displayed an entire want of care and a conscious and depraved indifference to the consequences of 16 its conduct, including to the health, safety, and welfare of their customers, and warrants an award 17 of punitive damages in an amount sufficient to punish each Defendant and deter others from like 18 conduct. 19 1020. Each of the Defendants is further liable to Plaintiffs and Consortium Plaintiffs for 20 punitive damages based upon its willful and wanton conduct toward underage users, including 21 Plaintiffs whom they knew would be seriously harmed using Defendants' respective social media 22 products. 23 COUNT 11: INTENTIONALLY OMITTED 24 25 1021–1038. Intentionally omitted. 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 263 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 270 of 278 1 COUNT 12: 2 VIOLATIONS OF 18 U.S.C. §§ 2255 and 2252 (Against Meta) 3 1039. Plaintiffs re-allege and incorporate by reference each preceding and succeeding 4 paragraph as though set forth fully at length herein. 5 1040. Plaintiffs bring claims under 18 U.S.C. § 2255 based on each Defendant's violation 6 of 18 U.S.C. § 2252 which prohibits the knowing receipt of a visual depiction of Plaintiffs engaging 7 in sexually explicit conduct while a minor. 8 1041. Each Defendant violated the federal child pornography crime found at 18 U.S.C. 9 § 2252(a)(1) which provides that any person commits a federal crime who: 10 (1) knowingly transports or ships using any means or facility of 11 interstate or foreign commerce […] any visual depiction, if— 12 (A) the producing of such visual depiction involves the use of a minor engaging in sexually explicit conduct; and 13 14 (B) such visual depiction is of such conduct. 15 1042. Each Defendant also violated the federal child pornography crime found at 16 18 U.S.C. § 2252(a)(2) which provides that any person commits a federal crime who: 17 (2) knowingly receives, or distributes, any visual depiction using any means or facility of interstate or foreign commerce […] 18 or knowingly reproduces any visual depiction for distribution 19 […] if— 20 (A) the producing of such visual depiction involves the use of a minor engaging in sexually explicit conduct; and 21 (B) such visual depiction is of such conduct. 22 1043. Plaintiffs suffered personal injury as a result of each Defendant's violation of 18 23 U.S.C. § 2252. 24 1044. 18 U.S.C. § 2255, entitled "Civil Remedy for Personal Injuries," provides that any 25 person who is a victim of a violation of 18 U.S.C. § 2252 and who suffers personal injury as a result 26 of such violation shall recover the actual damages such person sustains or liquidated damages in 27 the amount of $150,000 per victim, and reasonable attorney's fees. 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT -264- (PERSONAL INJURY) 2779607.1 CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 271 of 278 1 1045. Plaintiffs intend to prove actual damages as a result of each of the Defendants' 2 conduct. 3 1046. At a minimum, Plaintiffs intend to seek statutory liquidated damages in the amount 4 of $150,000 against each Defendant, as well as the cost of the action, including reasonable 5 attorney's fees and other litigation costs reasonably incurred, prejudgment and post-judgment 6 interest, and such other relief as the Court deems appropriate, pursuant to 18 U.S.C. § 2255(a). 7 1047. Each Defendant's conduct, as described above, was intentional, fraudulent, willful, 8 wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and displayed an 9 entire want of care and a conscious and depraved indifference to the consequences of its conduct, 10 including to the health, safety, and welfare of their customers, and warrants an award of punitive 11 damages in an amount sufficient to punish each of the Defendants and deter others from like 12 conduct. 13 1048. Plaintiffs demand judgment against each of Defendants for injunctive and monetary 14 relief, including compensatory, treble, and punitive damages, together with interest, costs of suit, 15 attorneys' fees, and all such other relief as the Court deems proper. 16 COUNT 13: INTENTIONALLY OMITTED 17 18 1049–1056. Intentionally omitted. 19 COUNT 14: VIOLATIONS OF 18 U.S.C. §§ 2255, and 2252A(5)(b) 20 (Against Meta) 21 1057. Plaintiffs re-allege and incorporate by reference each preceding and succeeding 22 paragraph as though set forth fully at length herein. 23 1058. Plaintiffs bring claims under 18 U.S.C. § 2255 based on each Defendant's violation 24 of 18 U.S.C. § 2252A(a)(5)(B) and each Defendant's knowing receipt of a visual depiction of 25 Plaintiffs while minors in violation of 18 U.S.C. § 2256. 26 1059. Each Defendant violated the federal child pornography crime found at 18 U.S.C. 27 § 2252A(a)(5)(B) which provides that any person commits a federal crime who: 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 265 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 272 of 278 1 knowingly possesses, or knowingly accesses with intent to view, any … material that contains an image of child pornography that 2 has been mailed, or shipped or transported using any means or facility of interstate or foreign commerce … or that was produced 3 using materials … affecting interstate or foreign commerce by any means, including by computer. 4 1060. Plaintiffs suffered personal injury because of each Defendant's violation of 18 5 U.S.C. § 2252A(a)(5)(B). 6 1061. 18 U.S.C. § 2252A(f)(2), entitled "Relief" provides that the court may appropriate 7 relief, including "temporary, preliminary, or permanent injunctive relief;" as well as "compensatory 8 and punitive damages; and the costs of the civil action and reasonable fees for attorneys and expert 9 witnesses." 10 1062. 18 U.S.C. § 2255, entitled "Civil Remedy for Personal Injuries," provides that any 11 person who is a victim of a violation of 18 U.S.C. § 2252 and who suffers personal injury as a result 12 of such violation shall recover the actual damages such person sustains or liquidated damages in 13 the amount of $150,000 per victim, and reasonable attorney's fees. 14 1063. Plaintiffs intend to prove actual damages as a result of each of the Defendants' 15 conduct. 16 1064. At minimum, Plaintiffs intend to seek statutory liquidated damages in the amount 17 of $150,000 against each Defendant, as well as the cost of the action, including reasonable 18 attorney's fees and other litigation costs reasonably incurred, prejudgment and post-judgment 19 interest, and such other relief as the Court deems appropriate, pursuant to both 18 U.S.C. §§ 2255(a) 20 and 2252A(f). 21 1065. Each Defendant's conduct, as described above, was intentional, fraudulent, willful, 22 wanton, reckless, malicious, fraudulent, oppressive, extreme, and outrageous, and displayed an 23 entire want of care and a conscious and depraved indifference to the consequences of its conduct, 24 including to the health, safety, and welfare of their customers, and warrants an award of punitive 25 damages in an amount sufficient to punish each of the Defendants and deter others from like 26 conduct. 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 266 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 273 of 278 1 1066. Plaintiffs demand judgment against each of the Defendants for injunctive and 2 monetary relief, including compensatory, treble, and punitive damages, together with interest, costs 3 of suit, attorneys' fees, and all such other relief as the Court deems proper. 4 COUNT 15: INTENTIONALLY OMITTED 5 6 1067–1080. Intentionally omitted. 7 COUNT 16: WRONGFUL DEATH 8 (Against All Defendants) 9 1081. Plaintiffs re-allege and incorporate by reference each preceding and succeeding 10 paragraph as though set forth fully at length herein. 11 1082. This Cause of Action is asserted by and on behalf of Plaintiffs bringing their actions 12 as heirs of Decedents or as duly-appointed representatives of the estates of Decedents pursuant to 13 the laws of various states. 14 1083. As a direct and proximate result of the conduct of each of the Defendants and the 15 defective nature of its respective social media products as outlined above, Decedents suffered 16 wrongful death, and Plaintiffs suing as heirs or estate representatives of Decedents seek damages 17 therefor, including loss of financial support, loss of society, funeral expenses, and estate 18 administration expenses as permitted under various states' laws, and where applicable punitive 19 damages 20 1084. Plaintiffs demand judgment against each of the Defendants for compensatory, 21 treble, and punitive damages, together with interest, costs of suit, attorneys' fees, as permitted under 22 various states' laws and all such other relief as the Court deems proper. 23 COUNT 17: SURVIVAL ACTION 24 (Against All Defendants) 25 1085. Plaintiffs re-allege and incorporate by reference each preceding and succeeding 26 paragraph as though set forth fully at length herein. 27 1086. This Cause of Action is asserted by and on behalf of heirs of Decedents or the duly- 28 appointed representatives of the estates of Decedents, pursuant to the laws of various states. PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 267 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 274 of 278 1 1087. As a direct and proximate result of the conduct of each of the Defendants and the 2 defective nature of its respective social media products as outlined above, Decedents suffered 3 bodily injury resulting in pre-death pain and suffering, disability, disfigurement, mental anguish, 4 emotional distress, loss of capacity of the enjoyment of life, a shortened life expectancy, expenses 5 for hospitalizations and other medical and nursing treatments, loss of earnings, and loss of ability 6 to earn. Plaintiffs suing as heirs or estate representatives seek damages for these injuries to their 7 respective Decedents as permitted under various states' laws, including where applicable punitive 8 damages. 9 1088. Plaintiffs demand judgment against each of the Defendants for compensatory, 10 treble, and punitive damages, together with interest, costs of suit, attorneys' fees, as permitted under 11 various states' law, and all such other relief as the Court deems proper. 12 COUNT 18: LOSS OF CONSORTIUM AND SOCIETY 13 (Against All Defendants) 14 1089. Consortium Plaintiffs re-allege and incorporate by reference each preceding and 15 succeeding paragraph as though set forth fully at length herein. 16 1090. As a direct and proximate result of the conduct of each of the Defendants and the 17 defective nature of its respective social media products as outlined above, the Consortium Plaintiffs 18 have necessarily paid and/or have become liable to pay, and will continue to pay and/or continue 19 to be liable to pay, for medical aid, medical treatment, and medications of the Plaintiffs in this 20 litigation. 21 1091. As a direct and proximate result of the conduct of each of the Defendants and the 22 defective nature of Defendants' respective social media products outlined above, the Consortium 23 Plaintiffs have been caused and will continue to be caused the loss of their childrens', wards', 24 spouses', parents', siblings', and/or other close family members' consortium, companionship, 25 services, society, love, and comforts, and their familial association has been altered, and, 26 accordingly, the Consortium Plaintiffs have been caused great mental anguish and emotional 27 distress. 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 268 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 275 of 278 1 1092. Each Defendant's conduct, as described above, was willful, wanton, reckless, 2 malicious, fraudulent, oppressive, extreme and outrageous, and displayed an entire want of care 3 and a conscious and depraved indifference to the consequences of its conduct, including to the 4 health, safety, and welfare of Plaintiffs, and warrants an award of punitive damages. 5 1093. Consortium Plaintiffs demand judgment against each of the Defendants for 6 compensatory, treble, and punitive damages, together with interest, costs of suit, attorneys' fees, 7 and all such other relief as the Court deems proper. 8 VII. PRAYER FOR RELIEF 9 Plaintiffs and Consortium Plaintiffs demand judgment against each of the Defendants to 10 the full extent of the law, including but not limited to: 11 1. judgment for Plaintiffs and Consortium Plaintiffs, and against each Defendant; 12 2. damages (both past and future) to compensate Plaintiffs and Consortium Plaintiffs 13 for injuries sustained as a result of the use of each Defendant's respective social 14 media products including, but not limited to physical pain and suffering, mental 15 anguish, loss of enjoyment of life, emotional distress, expenses for hospitalizations 16 and medical treatments, and other economic harm that includes but is not limited to 17 lost earnings and loss of earning capacity; 18 3. where alleged, damages to compensate Consortium Plaintiffs for loss of consortium, 19 companionship, services, society, love, and comforts, alteration to their marital or 20 filial association, and mental anguish and emotional distress; 21 4. where alleged, all damages available for wrongful death and survival; 22 5. exemplary, treble, and punitive damages in an amount in excess of the jurisdictional 23 limits; 24 6. attorneys' fees; 25 7. experts' fees; 26 8. costs of litigation; 27 9. pre-judgment and post-judgment interest at the lawful rate; 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 269 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 276 of 278 1 10. declaratory relief including, but not limited to, a declaration that each Defendant 2 defectively designed its respective products, failed to provide adequate warnings, 3 violated state consumer protection laws, and violated 18 U.S.C. §§ 1466A, 1595, 4 1591, 2252, 2252A(f), 2252A(5)(b), 2255, 2258A, and 2258B; 5 11. injunctive relief including, but not limited to, ordering each Defendant to stop the 6 harmful conduct alleged herein, remedy the unreasonably dangerous features in its 7 social media products, provide adequate warnings to minor users and parents that 8 its products are addictive and pose a clear and present danger to unsuspecting 9 minors, and prevent future violations of state law deceptive trade and practices acts, 10 including those premised on federal law such as COPPA; and 11 12. any other relief as this Court may deem equitable and just, or that may be available. 12 VIII. JURY DEMAND 13 Plaintiffs demand a trial by jury on all issues so triable. 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 270 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 277 of 278 1 Dated: December 15, 2023 Respectfully submitted, 2 /s/ Christopher A. Seeger /s/ Lexi J. Hazam CHRISTOPHER A. SEEGER LEXI J. HAZAM 3 SEEGER WEISS, LLP LIEFF CABRASER HEIMANN & 4 55 Challenger Road, 6th Floor BERNSTEIN, LLP Ridgefield Park, NJ 07660 275 Battery Street, 29th Floor 5 Telephone: 973-639-9100 San Francisco, CA 94111-3339 Facsimile: 973-679-8656 Telephone: 415-956-1000 6 cseeger@seegerweiss.com lhazam@lchb.com 7 /s/ Previn Warren 8 PREVIN WARREN MOTLEY RICE LLC 9 401 9th Street NW Suite 630 Washington DC 20004 10 T: 202-386-9610 11 pwarren@motleyrice.com Plaintiffs' Co-Lead Counsel 12 MICHAEL M. WEINKOWITZ JOSEPH G. VANZANDT LEVIN SEDRAN & BERMAN, BEASLEY ALLEN CROW METHVIN 13 LLP PORTIS & MILES, P.C. 510 Walnut Street, Suite 500 234 Commerce Street 14 Philadelphia, PA 19106 Montgomery, AL 36103 15 Telephone: 215-592-1500 Telephone: 334-269-2343 mweinkowitz@lfsbalw.com joseph.vanzandt@beasleyallen 16 EMILY C. JEFFCOTT JAYNE CONROY 17 MORGAN & MORGAN SIMMONS HANLY CONROY, LLC 220 W. Garden Street, 9th Floor 112 Madison Ave, 7th Floor 18 Pensacola, FL 32502 New York, NY 10016 19 Telephone: 850-316-9100 Telephone: 917-882-5522 ejeffcott@forthepeople.com jconroy@simmonsfirm.com 20 ANDRE MURA ALEXANDRA WALSH 21 GIBBS LAW GROUP, LLP WALSH LAW 1111 Broadway, Suite 2100 1050 Connecticut Ave, NW, Suite 500 22 Oakland, CA 94607 Washington D.C. 20036 23 Telephone: 510-350-9717 T: 202-780-3014 amm@classlawgroup.com awalsh@alexwalshlaw.com 24 MATTHEW BERGMAN 25 SOCIAL MEDIA VICTIMS LAW CENTER 26 821 Second Avenue, Suite 2100 27 Seattle, WA 98104 Telephone: 206-741-4862 28 matt@socialmediavictims.org Plaintiffs' Leadership Committee PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 271 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047 Case 4:22-md-03047-YGR Document 494 Filed 12/15/23 Page 278 of 278 1 RON AUSTIN PAIGE BOLDT 2 RON AUSTIN LAW WATTS GUERRA LLP 400 Manhattan Blvd. 4 Dominion Drive, Bldg. 3, Suite 100 3 Harvey LA, 70058 San Antonio, TX 78257 4 Telephone: (504) 227–8100 T: 210-448-0500 raustin@ronaustinlaw.com PBoldt@WattsGuerra.com 5 JAMES J. BILSBORROW SIN-TING MARY LIU 6 WEITZ & LUXENBERG, PC AYLSTOCK WITKIN KREIS & 7 700 BROADWAY OVERHOLTZ, PLLC NEW YORK, NY 10003 17 East Main Street, Suite 200 8 Telephone: 212-558-5500 Pensacola, Fl 32502 jbilsborrow@weitzlux.com Telephone: 510-698-9566 9 mliu@awkolaw.com CARRIE GOLDBERG 10 C.A. GOLDBERG, PLLC ROLAND TELLIS 11 16 Court St. BARON & BUDD, P.C. Brooklyn, NY 11241 15910 Ventura Boulevard, Suite 1600 12 T: (646) 666-8908 Encino, CA 91436 carrie@cagoldberglaw.com Telephone: (818) 839-2333 13 rtellis@baronbudd.com EMMIE PAULOS 14 LEVIN PAPANTONIO HILLARY NAPPI 15 RAFFERTY HACH & ROSE LLP 316 South Baylen Street, Suite 600 112 Madison Avenue, 10th Floor 16 Pensacola, FL 32502 New York, New York 10016 Telephone: 850-435-7107 Tel: 212.213.8311 17 epaulos@levinlaw.com hnappi@hrsclaw.com 18 DIANDRA "FU" DEBROSSE JAMES MARSH 19 ZIMMERMANN MARSH LAW FIRM PLLC DICELLO LEVITT 31 Hudson Yards, 11th Floor 20 505 20th St North, Suite 1500 New York, NY 10001-2170 Birmingham, Alabama 35203 Telephone: 212-372-3030 21 Telephone: 205.855.5700 jamesmarsh@marshlaw.com fu@dicellolevitt.com 22 23 RUTH RIZKALLA CARLSON LAW FIRM 24 100 E. Central Texas Expy Killeen, TX 76541 25 (254) 526-5688 RRizkalla@carlsonattorneys.com Plaintiffs' Steering Committee 26 27 28 PLAINTIFFS' SECOND AMENDED MASTER COMPLAINT - 272 - (PERSONAL INJURY) CASE NO. 4:22-MD-03047

Plaintiffs' Second Amended Master Complaint (Personal Injury)

Raw text