2014–2026
539 passages quoted from Meta's internal records, court filings and congressional testimony, in date order. Each is labelled with its source.
PLAINTIFFS’ CHARACTERIZATION
Likes
Direct Messaging
"Touchpoints Leveraging IG: Help Kids Build Up to Social with Adjusted IG Features" "Content Relevant to Youth: Onboard brands such as Barbie, Paw's Control, BabyBus, and Toca Boca onto IG Tweens (most tween appropriate brands are not currently on IG). Invest in youth relevant content by specific age bands (e.g. Onboard Youtube creators of UGC children's shows, partner with kids content brands/publishers)" "Segregated Age Experience: Ages <5: Reels or IGTV for kids. Social is a passive social experience to kids at this stage. They are interested in learning about their friends but not active in engagement and creation. Ages 5-7: Reels, Feeds, Profile, and Explore. Older pre-tweens are starting to have some preferences and desire to create. Surfaces that allow creation and consumption of others' creation can be introduced with appropriate content. Ages 8-12: More private sharing. As tweens age with growing need of socialization, DM and Stories functions can also be introduced, focusing on close friends, unique expression, shared close friends experience" "Relevant Expressions: Esp for tweens: Creative and intuitive tools to teach kids how to edit photos, and create art, stickers, and effects; kids can also be developers"
Bullying & harassment
Bryce Jon Bartlett: "I think there is a tradeoff in B&H with other values (like voice, creation, and community). I think that a successful vision is more about transparency, fairness, and balancing competing values than about eliminating B&H or offensive content." Bryce Jon Bartlett: "I learned some teams have a 'budget' for integrity regressions. I.e. over a half, they can launch products that regress integrity but not more than XX in aggregate over a half." Eden Esther ████: "...though we have held back on certain launches on gaming per request from the gaming team since certain behaviors are much more acceptable within the gaming context and we were seeing larger engagement drops there."
Child sexual exploitation
████: "#wut PBD" ████: "Private by default :)"| ████: "Ok got it… Sorry for being a dumb n00b, but boosting is what, having new teens content ranked higher so they get more engagement or am I way off?" ████ it was about boosting teens appearances in Suggested Users. WHich is something growth is testing now for US Android in general" ████: "I'm still kinda confused about the goal. We want to make sure teens don't accidentally share more than they want, and avoid creepers, but then some of them totally want that and then if we boost them, we are inviting people to see more teens in their suggestions. Is it about decoupling clear privacy control from integrity issues, or is it about saying we want to reduce integrity problems, but not wanting the consequence of less interactions and follows?" ████: "It is about looking good to regulators so that they don't block our under 13 year old IG version we are working on. Thats it. It has a terrible impact on teen engagement and retention and no detectable benefit on integrity metrics."
PLAINTIFFS’ CHARACTERIZATION
Content Algorithm
PLAINTIFFS’ CHARACTERIZATION
Likes
"During the initial testing of Daisy controls there were two approaches to the default setting. While some teens had the choice to hide their like counts (opt-in condition), others had their like counts hidden for a period of time and then were given the option to opt-out of Daisy controls (opt-out condition)." "when daisy controls are opt-in, only 0.72% of people choose to hide like counts, but when they're opt-out, 35% leave their like counts hidden." "In the opt-out condition, 69% of teens were aware of Daisy controls as a product. This was significantly higher than the 49% of teens in the opt-in condition who were aware of Daisy." "Teens in the opt-out condition were significantly less likely to feel worse about themselves because of the number of likes other people received on their Instagram posts. This is one of our most relevant indicators of negative social comparison and demonstrates the impact default Daisy controls can have on experiences of comparison on the platform."
AutoplayInfinite Scroll
████: "Most relevant Google/Youtube Well-being changes to our team... 'On YouTube, we'll turn on take a break and bedtime reminders and turn off autoplay for users under 18.'" ████: "Turning off autoplay for teens seems like a huge move!" ████: "Imagine if we turned off infinite scroll for teens" ████: "Yeah, I was thinking the same thing! Autoplay is HUGE!" ████: "Yeah, we should get more in touch with Youth Platform and IG Youth teams, since I think some mechanics changes have been discussed."
Addiction & problematic useBullying & harassmentSuicide & self-injuryFriend/Account Suggestions
"Mark KX, Pratiti and ████ have put together the proposal below for additional investment to strengthen our position on wellbeing across the company. Chris, Alex and I strongly endorse this approach, and I wanted you to see their thinking ahead of the discussions on headcount allocation." "From a Policy perspective, this work has become increasingly urgent over recent months. Politicians in the US, UK, EU and Australia are publicly and privately expressing concerns about the impact of our products on young people's mental health. In the US, this was specifically raised with me by the Surgeon General, and is the subject of potential legal action from State AGs. We have received numerous policymaker inquiries and hearing requests." "[W]e need to do more and we are being held back by a lack of investment on the product side which means that we're not able to make changes and innovations at the pace required to be responsive to policymaker concerns. Our wellbeing work is both under-staffed and fragmented across teams, which is why we view the proposed investment outlined below as important." "We are not on track to succeed for our core well-being topics (problematic use, bullying & harassment, connections, and SSI),* and are at increased regulatory risk and external criticism. These affect everyone, especially Youth and Creators; if not addressed, these will follow us into the Metaverse."
Body image & eating disorders
Stephanie ████: "Jeff Horwitz has a leaked social comparison research deck from 2020, he's writing a story that essentially argues that IG's design is inherently bad for teenage girls (leads to SSI, poor mental health, dysmorphia) and is not something we can fix with product, or an app design overhaul" Stephanie ████: "main angle: IG is bad for teen girls according to our own research, leads to poor mental health outcomes, our app is inherently flawed for vulnerable groups" Adam Mosseri: "I'm sure it's too nuanced, but it's saying it will examine the connection" Stephanie ████: "I'm mostly worried about the fallout from the article, moreso than pull quotes - that our own research confirmed what everyone has long suspected"
████: "I'm not sure if folx have reached out to you about this deck from you and ████ last year." ████: "It was unfortunately leaked to a reporter and there are some questions coming up about some of the insights stated in there" ████: "Yeah...the reporter reached out to me, too" ████: "but wasn't sure until today that it was that deck that got leaked" ████: "just suspicious because the reporter mentioned teen girls and that's the only work I've done exclusive with that demo" ████: "yeah. in addition to some of the questions coming up, we're also locking down access to some of the extra sensitive pieces of work" ████: "This may have already been flagged to you (outside the mental health deck I mentioned above), but can you help make sure that any of our shareable deliverables or insights docs that you own on the mental well-being space are locked down (so that people have to be specifically added to the docs - link share isn't on)?" ████: "no, no one mentioned that to me. Will do it now" ████: "ok I've also added you to a doc where we're going to try to pull together the high level summaries of research we've done in the social comparison/ mental health-well-being spaces as well"
Dayna Geldwert: "I am so devastated" Dayna Geldwert: "We have to pull back on all our comms / efforts for Girl Scouts" ████: "omg what... why" Dayna Geldwert: "Strategic Response said we can't given the leak" ████: "i don't understand why!?" Dayna Geldwert: "Their sense is would seem strange/reactive/not land well to go hard on how we're supporting girls just days after a very damning article comes out about how we are damaging the well-being of girls."
Addiction & problematic use
████: "do you happen to have anyyyy screenshots of recent [TikTok]] take a break?? or know how long i need to scroll for to find it lol" ████: "i'm trying to make a case for why we can't water down our Take a Break" ████: "so i want to show Miki etc what TT already does" ████: "ughh they've already done everything we were brainstorming lol" ████: "████ is like so concerned about us telling people to leave the app" ████: he's not helping us w anything right now." Dayna Geldwert: "That's so annoying" Dayna Geldwert: "Telling them to leave and forcing them to leave is different" ████: "exactly" ████: "i think what we'd do differently is allow people to customize -- choose after how long they see the breaks etc" ████: "but he wants this to be OPT IN and im like nO" Dayna Geldwert: "Ok but we keep making tools no one is using" Dayna Geldwert: "because we make it a control" Dayna Geldwert: "soooooooo" ████: "ya like we have to at least show ppl!!!!!" ████: "maybe we could show it like 3 times and stop showing if they don't opt in" ████: "but no one will" ████: "i said do it for teens at least!" Dayna Geldwert: "Ugh"
PLAINTIFFS’ CHARACTERIZATION
PLAINTIFFS’ CHARACTERIZATION
Addiction & problematic use
PLAINTIFFS’ CHARACTERIZATION
"Our overall growth in the US has plateaued and we've recently seen concerning decline in US Teen growth, especially with early teens, indicating an existential threat for not just IG but the company overall. The Growth team's top goal this half is to reverse this decline and re-accelerate US growth."
Body image & eating disorders
"A lot of NSC has been attributed to celebrities and influencers, and we've currently been looking at controls and ranking to solve this." "However, it's not an influencer problem, it's a culture problem. Those influencers might be responding to Instagram's culture and norms by filling their Instagram profiles with NSC triggering content in a way that doesn't seem to be happening on TikTok or other platforms." "Thus, controls and ranking are a starting point but not a final solution. Ultimately, winning requires IG to reshape its culture and norms." "Negative social comparison (NSC, defined in our people problems) is a significant problem on Instagram, with 68% of teen girls experiencing negative social comparison on Instagram and users saying that Instagram exacerbates the problem more than any other mental health problem. Media, policy partners, and governments have also been vocal about Instagram's role in encouraging social comparison that can lead to anxiety, stress, and depression." "While NSC is not unique to Instagram, users feel that NSC is more common on (also see the quant results) than any competitors or the Facebook app. In an early 2020 IG Mental Health user study, 15 participants ranked IG in its own separate tier for most NSC"
Body image & eating disordersContent Algorithm
"It's not an influencer problem, it's an Instagram problem. It's not a format problem, it's a culture problem. The very same celebrities and influencers who feel comfortable sharing their family, bloopers, highlights, and lowlights on TikTok respond to Instagram's culture and norms by filling their Instagram profiles with NSC triggering content." ████: "We decide, via ranking, what kinds of posts are promoted to millions of people in Explore. If we showed less NSC-triggering content to teens in Explore, that would change teens' impression of what kinds of posts Instagram is for, and would change the kind of content that creators produce because their overly-perfect content wouldn't perform as well as their more fun, authentic, personal content. Currently, we've chosen to have no explicit POV on NSC-triggering content, and as far as we can tell we're actively recommending a LOT of it to teens, especially teen girls. Having a POV on NSC-triggering content and incorporating it into ranking on our recommendations surfaces seems like a clear opportunity to nudge Instagram's culture and change both what's shown to teens and what's produced." ████: "I also think there is something very key here -- solutions for social comparison are pivotal if we want a strategy for youth to land."
Child sexual exploitationContent Algorithm
"Reels featuring kids/tweens, and teens (i.e., 0-15) dancing and posing are pervasive, and often show up on people's Reels experience, even in cold starts. While much of this content is benign (they may be uninteresting to users, but not necessarily harmful), some may include risky elements (i.e., partial nudity, suggestive dance moves that are borderline CEI) that make both viewers uncomfortable and the subjects at risk." "Sexually suggestive content featuring minors (SSCM) (labeled prevalence) is 2-3x more prevalent on Reels home than on Explore." "Female and unknown gender teens are producing more content marked sexually suggestive on Reels, potentially opening them up to greater risk. According to both reporting and prevalence data, female and unknown gender teens create more, while male teens report more SSCM content." "We need to better detect and filter content depicting minors in sexually suggestive ways on Reels, as the reporting vs. prevalence data suggests that we're permitting many to "slip through the cracks". Improving our detection of minors, as well as getting ahead of problematic trends (i.e., upcoming dance trends that may put children at risk) may help us improve." "Benign but still risky content featuring minors: Children dancing/posing unto itself is not problematic, but it becomes an issue if bad actors see it. New approaches to protect children and teens (i.e., preventing visibility between known IIC adults and minors on other surfaces like Explore) should be ported to Reels as soon as possible."
Child sexual exploitation
████: "before my time (when covid hit), the decision was made to prioritise our (tiny) amount of review capacity for English only, FB only… which means 1% of total report. Our goal this half it to be able to deal with 100%... and i'm pretty shocked i had to fight for resource to close that gap" ████: "i'm getting a bit tired of us being traded off against far more mild harms. I feel like all i do is escalate" ████: "look at this - i had to deprioritise goals AGAIN today because we have such a massive codebase to maintain" Rothschild: "How big is your team? It could be that the baseline resourcing needs some revisiting such that you don't have to constantly escalate." ████: "5 [engineers], 1 open head + 2[Data Scientists], 1[Data Engineer]" Rothschild: "And this is to cover all child safety problems? I.e. IIC [Inappropriate Interactions with Children], CEI [Child Exploitative Imagery] etc?" ████: "plus the multiple other problems that i have to 'non goal' on… no sexual child abuse… child nudity… which means i practice we have like 1 engineer for all of IIC, one for all of CEI etc… which feels pretty wild to me."
DrugsAddiction & problematic useContent Algorithm
Miki Rothschild: "[W]ith the recent PR fire TikTok has seen around inappropriate content for teens (see here), we were chatting within Youth XFN about our own deficits when it comes to discoverability, namely: (1) it's still very easy to find borderline and sometimes violating content/accounts in search, e.g. try typing 'drug' or 'love' (2) if you start following borderline accounts or interacting with such content, our recommendations algorithms will start pushing you down a rabbit hole of more egregious content." "3. How are we thinking about non-rec/violating Rabbit holes in recommendations? Do we have any specific approaches to this?" ████: "On #3 we've done some early experimentation with the Drebbel team on additional enforcement for users who we detect to be in a rabbithole on Feed Recs. TL;DR we did see a meaningful drop in exposure with targeted demotion, but it came with a clear engagement cost and the effects did not last after intervention."
Arturo Bejar: "It's been almost two years since I started supporting the Well-being team at Instagram. During that time and through my conversations with people I have found what, in my experience, is the most important thing for Instagram to be working on. I think it is something that needs your attention and leadership." Arturo Bejar: "I know that this week has very busy, and apologize for opening a difficult topic during this time. I wanted to add some context. In Well-being, a researcher named Kyle Andrews is developing a new measurement tool, building on top of TRIPS to help our understanding of bad experiences. As part of that he did an age cut, where you can see the reach of experience on the issues surveyed, the cells is red are where an issue presents the most in the last 7 days in a particular age range. The 13-15 year range numbers are worth your attention, and highlight the importance of our understanding of bad experiences on the youth work." Adam Mosseri: "Arturo, I'm pretty underwater right now with a few big things but will try and read this this week. Thanks for sending."
Nick Clegg (9/14/2021, forwarding the story to Alex Schultz and Pratiti Raychoudhury): "...this is going to feed directly into ongoing regulatory/legislative actions in UK (On Line Harms Bill), US, EU and elsewhere. My guess is of the barrage of stories that the WSJ is running about us this week, this one will be most impactful." Pratiti Raychoudhury (9/20/2021): "Nick -- first of all I wanted to thank you for the newsroom post. It really lifted the morale of the (internal) research community. I know I am preaching to the choir, but given last week's events, I feel even more convinced that we need to make more progress on well-being on the product side. From everything I am hearing (second hand) that there will be no headcount. The plan that the team had put together is bare minimum to get us to parity with TikTok. And now TikTok has gone even more aggressive [link to TechCrunch article on TikTok expanding mental-health resources]. I know you have been advocating for this. So wanted to put this on your radar." Nick Clegg (9/20/2021): "Thx Pratiti -- under whose HC ask would this have been included? It's not in mine, right? I had to leave the SG discussion mid way to catch a flight -- I think we're aligned on 3a for internal research? How about external?" Pratiti Raychoudhury (9/20/2021): "It will be under Javi/Naomi. But your advocacy will go a long way here! Yes, we netted out with 3a for internal research. We had only about 5 mins for external research and he wanted to spend more time on this. Alex made a case for doing at least the well-being (external) work with OII." Nick Clegg (9/21/2021): "Ok, and what's the timing of the H/C cycle -- I think the 'napkin' is more or less locked, isn't it? Do you have a specific ask for Javi/Naomi that I could resurface?" Pratiti Raychoudhury (9/21/2021): "We haven't received headcount numbers yet -- ████ and ████ will follow up with you on the next steps"
Pratiti Raychoudhury: "I know I am preaching to the choir, but given last week's events, I feel even more convinced that we need to make more progress on well-being on the product side. From everything I am hearing (second hand) that there will be no headcount. The plan that the team had put together is bare minimum to get us to parity with TikTok. And now TikTok has gone even more aggressive... I know you have been advocating for this. So wanted to put this on your radar." Nick Clegg: "Thx Pratiti -- under whose HC ask would this have been included? It's not in mine, right?... I think we're aligned on 3a for internal research? How about external?" Pratiti Raychoudhury: "It will be under Javi/Naomi. But your advocacy will go a long way here! Yes, we netted out with 3a for internal research. We had only about 5 mins for external research and he wanted to spend more time on this. Alex made a case for doing at least the well-being (external) work with OII." Nick Clegg (9/20/2021): "Ok, and what's the timing of the H/C cycle - I think the 'napkin' is more or less locked, isn't it? Do you have a specific ask for Javi/Naomi that I could resurface?" Pratiti Raychoudhury (9/21/2021): "We haven't received headcount numbers yet - and ████ and ████ will follow up with you on the next steps"
Sources include internal documents, public filings and press reports. “Alleged by plaintiffs” denotes the plaintiffs’ characterization, not Meta’s own words. Source labels identify provenance, not a finding of fact.
Meta & Children's Online Safety