Child Safety: State of Play

Raw text

Transcribed text as recorded, including headers and formatting fragments.

Exhibit 91 Message r' I I P" h r' Ir 3 i: From: [/0=THEFACEBOOK/OU=EXCHANGE ADMINISTRATIVE GROUP (FYDIBOHF23SPDLT)/CN=RECIPIENTS/CN-OZ] 20?" JAN 29 PH 5 09 Sent: 10/17/202012:15:27 AM To Nickc'ege 2.; Subject: Antigone doc DA'v'IDSL/'li CO. CIIAHCERY CT Attachments: Rethinking Child Safety AD .pth Z 'All __.D C. 8; H. Hi Nick Here are the slides Antigone has put together. There is some good material here on the scale of the internal problem illustrating why we're just not set up adequately to answer the challenge. I think she could both unpack some of those problems (slides aren't particularly conducive to the level of detail that's needed here), and also expand on the section near the beginning to emphasize the cumulative regulatory pressures we're under, and why they create both inevitability and urgency. But thought i would get it to you now so you can provide steers. I (FYI don't know what all the acronyms are on slide 9 I've asked Antigone to write them out). I CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT- IG-AG-00699249 . A)! R 11 J. 'i .9"!'1: u P . .u . m u u ,. L' 2 a .'I n ., In. 'v.' 1:: Vi v' '0 m i M». . w '33 {if L'I '4. 4 'If.' w l v}. ,w I . WK ,, ;. H } ' [4' it .- V: 'H '3 n.. ' '4; ,. 'h . -.o .u 1.». ,_ n q. 1. M» 5*- "'1' .. '5': 'sw J: ,_. 4v \ . '- "M 'k. >L '4 'flf . r,' z'» "Y .1 I. i:A. .3" .. 7 x: u ., £43 . ; E an; v. ,a ""h ;~« qfi m A" 1n - ,. L") n .. c fi' '1'. ,.4 an. d. U .( '14 U "In; '. J "'1 w. .' Ll. g. 1. wu m l,' .u .J-v gr K, so 11 a! v_t 5' '9» .. '. h Nu '5' g5 1.. 4.. w . H? U1, E MW -,. .. .13; '1' m 9-5. 1.. (A .. . m'" .d '1' V. 43%-..6 my:'9 -, h.» ' V -.5 n =1 5*; .m .. 'vv .'1 p 'f. rf': -.u. far' I'M.'i,-.w b r. -.1 v. .u '{ -L '1',"1'; 41ml , ,. 314. Wm. ' '1 a. r.' ,by" I, .32, CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT-IG-AG-00699250 3-'-r:* F'JL :\ Child Safety 1. Policy Landscape 2. State of Play 3. Proposal CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT-IG-AG-00699251 Child Safety fueling existential policy risks l9 Child safety enables policymakers and advocates to focus the public on visceral issues and ignore complexities. We need the complexities to be part of the discussion if we hope to move toward reasonable legislation across important areas like content regulation, age of consent, and user privacy. Addressing child safety in a more robust way will NOT clear away legislative efforts in these areas, but it WlLL enable us to guide legislation toward models that align more closely with our legislative and business goals (e.g., content regulatory model that looks more like what has been proposed by CCIA than what we are seeing w/EARN IT or CornynMachin or worst versions of UK Online Harms). CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT-IG-AG-00699252 Child Sexual Exploitation State of Play Vulnerabilities | Opportunities • Data: under resourced, just unlocking valuable insights • User education: minimal • Growth: not resourced to address growth concerns with adjust to mitigations like forward and group size limitations • Make significant investment in x-industry/multi-stakeholder initiative to improve ecosystem • Build highly effective prevention based system • Build crime victim resource hub • Offer legislative/regulatory model • Develop new success metrics Key point for slides 4-8: Not only do we need to solve for significant vulnerabilities but there are some important policy, user engagement and business opportunities we can unlock. CONFIDENTIAL - PROTECTED BY CONFIDENTIALITY AGREEMENT MT-IG-AG-00699253 Age Assurance/Appropriateness State of Play Vulnerabilities | Opportunities • U13 enforcement: lack classifier, no way to detect and remove at scale, U13 700k enforcement backlog and poor cross app enforcement, deprioritized enforcement due to concern re. legal liabilities • U18 enforcement: age gating relies on either stated age or weak age models, lack checkpoint • Age verification pushed to device, slow advancement, dependent on x-industry cooperation • Age models: not very accurate, not globally resilient and not globally operational, weak models negatively impact other integrity measures, including safety classifiers, uniquely poor efficacy on IG • Defaults, gating rules inconsistent, need wholesale review • Minors in-product education: minimal • Content: inappropriate/harmful content and experiences for minors • Age appropriate experiences • Unlock interoperable aging up (e.g., MKids to tween experience to IG & FB) • Device based age verification Note: There is work happening in this area but not resourced to move quickly CONFIDENTIAL - PROTECTED BY CONFIDENTIALITY AGREEMENT MT-IG-AG-00699254 SSI State of Play Vulnerabilities | Opportunities • Classifiers: technical, legal and operational challenges to global use • Borderline and aggregated content: lack classifiers to address at scale • Resources: lack differentiated responses and resources recommended by experts • Privacy: infrastructure limitations inhibit ability to make necessary guarantees • IG: DMs, groups being used to promote SSI and we don't scan there; and have no e2e resilient tech yet • Build Global text-based crisis response system • Launch x-industry info sharing via threat exchange • Develop industry recommended practices CONFIDENTIAL - PROTECTED BY CONFIDENTIALITY AGREEMENT MT-IG-AG-00699255 Well-being State of Play Vulnerabilities | Opportunities • Research: lack of good research; definitional, longitudinal, causation v correlation • Product: core product features connect to challenging societal issues ◦ objectification of women (e.g., AR face and body altering filters) ◦ competitive social comparison (e.g., likes and comments) ◦ anxiety/FOMO (e.g., notifications) • Tools: gaps in bullying and harassment tools (e.g., comment moderation and IG DM controls) • Policies: calibrating for content impact on well-being (e.g., eating disorder content and gender based hate speech) • Global text-based mental health support system • Tele mental-health • X-Industry digital well-being lab CONFIDENTIAL - PROTECTED BY CONFIDENTIALITY AGREEMENT MT-IG-AG-00699256 Proposals / 1 1 1 l i 1 i': , . 'v1[ -1 1 i ':1 e structure: oar-views to success, Child safety htegiity sits 'nsido of broader integrity team meaning "one of our most importam" responsibilities is constantly prioritized against a W'dE' swath of integrity issues, creating almcs: impossible choices. Additionall'y ow current approach has created a barrier to youth centerfid business OUUC'IU' 7'. es x r 'I' i Bil-(ids as a case study On the the Youth Team option, it is worth noting the engineering director already has mapped an interop age up strategy that integrates child safety, seeing it as core to unlocking product opportunities backed by research... To build a Youth Team, you do not necessarily need to hire tons of new headcount. You would need a strong executive lead who would use a model like KX had done for Health. CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT-IG-AG-00699257 The Numbers Child Safety | Central Privacy | Health • 1 PM + 6 ENG + 1 TBH, CI • 9, CO • 1, DS • 2, Policy (*Policy has a number of folks doing 25-5 time on Child Safety) • 7, OS • 615 • 175 TBH • 106 CW • 38 Eng • 9 PMs • 10 Partnerships • 1 + 2 TBH, Policy (*Policy has a number of folks doing 25-5 time on Child Safety) CONFIDENTIAL - PROTECTED BY CONFIDENTIALITY AGREEMENT MT-IG-AG-00699258 Central Privacy Structure B O W "mm x 33 O B «mega(Tu. 9 mm w n lawn-u lama-7mm 9 Wuhan « w w 0 vim-m v u w mm Key Features: Senior Product Executive Sponsorship VP leadership across infra, eng, strategy and implementation Inclusion of policy within the product team Has scope and growth, which should help get better and bigger hiring pipeline CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT-IG-AG-00699259 Health Structc Key Features: Central Senior Product Manager, who heads a Central XFN team and reports directly to Senior Product Executive (Naomi Gleit) XFN team cuts across strategy through to implementation XFN team has leads for each vertical dedicated solely to health Links integrity and opportunity, by integrating business goals could potentially unlock youth products Has scope and growth, which should help get better and bigger hiring pipeline CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MTlG-AG-00699260 {1:3 m lien-:3 Pg; x. '5 o Senior executive sponsorship Central senior product lead dedicated solely to relevant field/issue IC-7 to VP leadership across strategy. infra. eng and implementation tea ms; most dedicated solely to relevant field/issue Policy imbeddcd in central team CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT-lG-AG-00699261 Lit?» L Discussion 1. Do we agree we are not properly resourced to address child safety issues? 2. Do we think it is worth more deeply exploring an org structure shift? Does either model presented seem like the right starting point? 3. While we do that should we move fon/vard with more HC for existing teams with child safety focus? CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT-lG-AG-00699262 CICEDOO w « , CONFIDENTIAL PROTECTED BY CONFIDENTIALITY AGREEMENT - MT-IG-AG-OO699263