Big Tech and the Online Child Sexual Exploitation Crisis (QFR Responses)

QuestionWhy did Zuckerberg reject his own staff's 2021 recommendations (including from Nick Clegg) to bolster resources addressing problematic teen platform use, before cutting the Well-Being and Responsible Innovation teams?

ResponseMeta denies rejecting or ignoring the Well-Being Team's concerns, pointing to features developed since (Take a Break, Quiet Mode, Nudges, parental supervision tools) and overall safety spending, without directly addressing the Responsible Innovation Team's disbandment or the staff cuts described in the question.

QuestionWhat accountability should Meta face if it under-resources efforts to protect children from online child sexual exploitation?

ResponseMeta cites its overall safety investment and NCMEC praise rather than proposing any specific accountability mechanism for under-resourcing child protection efforts.

QuestionWhy did Meta reverse its 2022 opposition to app-store gatekeeper power and now support app-store-level parental approval for teen app downloads?

ResponseMeta says its 2022 NTIA comments and current app-store-parental-approval position are consistent, framing both as concerns about Apple's market dominance, and adds that any youth-safety legislation should include antitrust-style safeguards against Apple self-preferencing.

QuestionGiven a Stanford study finding CSAM in a popular AI training dataset, what is Meta doing to ensure Llama 3 and its other generative AI models aren't trained on datasets containing CSAM?

ResponseMeta describes model training to reject exploitative queries, continual red-team testing, output-scanning to block exploitative content, and user feedback tools, but does not directly confirm whether its own training datasets were screened for CSAM or address the Stanford dataset finding specifically.

QuestionGiven that disappearing-message features have made Snapchat a hub for sextortion, what is Meta doing to prevent its own disappearing-message features (vanish mode, Messenger) from putting more kids at risk of sextortion?

ResponseMeta lists its anti-sextortion measures (policies, financial-sextortion detection automation, restricted adult-teen messaging defaults, Take It Down/StopNCII partnerships with NCMEC, screenshot notifications) but defends keeping the disappearing-message feature itself rather than describing new safeguards specific to that feature's sextortion risk.

QuestionPlease itemize the 30+ parent/teen safety tools Meta has introduced over the past 8 years, including launch dates and efficacy assessments.

ResponseMeta provides a 2010-2024 chronological list of 50+ features/tools (comment filters, Restrict, Sensitive Content Control, Quiet Mode, Take a Break, parental supervision tools across apps, Take It Down) but does not describe how any of them were assessed for efficacy before or after launch, aside from generic 30-day retention statistics at the end.

QuestionAsked whether Meta supports the EARN IT Act (S. 1207) and why or why not.

ResponseMeta says it would generally support child-safety industry standards legislation but insists any such bill must protect encryption and not undermine Section 230.

QuestionAsked what measures Meta takes to prevent and detect sextortion, including financial sextortion, in real time.

ResponseMeta describes a multi-pronged sextortion approach: policy bans, specialized detection teams, NCMEC referrals, default private/restricted-messaging settings for teens, and partnerships like Take It Down/StopNCII, while calling on Congress for app-store-level age verification.

QuestionAsked whether Meta seeks feedback from CSAM/sexual-exploitation survivors to improve trust-and-safety policy, with examples.

ResponseMeta says it consults survivors and survivor organizations, funds NCMEC survivor services, and works with child-safety professionals to track evolving coded language used to evade detection.

QuestionAsked for examples of Meta's collaboration with parents and parent organizations on child online safety.

ResponseMeta lists parent-facing programs (Family Center, Messenger Kids advisory input, TTC Labs, Screen Smart events) and researcher partnerships (Digital Wellness Lab, Aspen Institute) that informed features like Take a Break and Quiet Mode.

QuestionAsked why Meta sets its minimum sign-up age at 13, and why not younger or older.

ResponseMeta says the age-13 minimum follows COPPA and matches industry-standard practice.

QuestionAsked how many minors use Meta's platforms and how much annual revenue Meta makes from them.

ResponseMeta gives self-reported teen-user percentages for Instagram and Facebook but does not answer the revenue question at all.

QuestionSub-part of Question 11 asking whether Meta has tested or is developing technology to detect CSAM in live video.

ResponseMeta cites existing photo/video-matching tools (PDQ, TMK-PDQF, PhotoDNA, Google Content Safety API) and a September 2023 automated enforcement effort that increased automated deletions of nudity/sexual-activity Instagram Lives fivefold, without directly confirming dedicated live-video CSAM detection R&D.

QuestionAsked what metrics Meta uses to measure the effectiveness of its child sexual abuse/exploitation trust-and-safety efforts.

ResponseMeta says it uses "prevalence" (viewership of violating content) as its key metric, estimating CSAM prevalence on Instagram at under 0.01%.

QuestionAsked whether Meta uses language-analysis tools to detect grooming and what further investments it will make.

ResponseMeta confirms it uses language-analysis tools and behavioral signals to flag potentially suspicious adult accounts and deploys search interstitials, without directly stating whether these are used to detect grooming conversations themselves.

QuestionAsked what resources Meta has built for victims/survivors of abuse on its platforms.

ResponseMeta lists over 50 victim/survivor-facing tools (Restrict, anonymous reporting), NCMEC/Take It Down/StopNCII partnerships, Thorn-developed sextortion guidance, the new nudity-protection DM feature, and industry collaborations (Technology Coalition, Project Protect, Lantern).

QuestionAsked how Meta responds to content-removal requests from CSAM survivors and the public.

ResponseMeta says it removes and reports CSAM to NCMEC, maintains a dedicated survivor-flagging channel, and points to Take It Down/StopNCII and 24/7 reporting review as its response mechanisms.

QuestionAsked how Meta will address CSAM becoming invisible to it once Messenger is end-to-end encrypted, and whether it has estimated the resulting percentage drop in CSAM detection/reporting to NCMEC.

ResponseMeta describes non-message-content signals (behavioral detection, on-device nudity protection) it will still use post-encryption and cites WhatsApp's CSAM reporting volume as reassurance, but does not give the requested percentage-reduction estimate.

QuestionAsked how Meta will prioritize the privacy of children/CSAM survivors whose abuse imagery continues to circulate, given the shift to default E2EE on Messenger.

ResponseMeta repeats its three-pillar safety approach for encrypted messaging (prevent, control, respond quickly), citing default private teen accounts, restricted DM settings, the new nudity-protection feature, and industry coalitions, but does not directly address CSAM survivors' ongoing exposure.

QuestionAsked what specific technical investments Meta will make to detect CSAM within end-to-end encrypted environments.

ResponseMeta declines to give a distinct answer, deferring entirely to its prior responses on Questions 16 and 17.

QuestionAsked Zuckerberg to reconcile his hearing testimony citing a National Academies report with that same report's call for further research and a national industry standard.

ResponseMeta defends its testimony by quoting the National Academies study's population-level findings and points to funded research efforts (NIH, Center for Open Science, Oxford Internet Institute) as evidence it supports further study.

QuestionSub-part of Question 19 asking whether Meta supports a national CSAM industry standard and, if so, why it does not support the EARN IT Act.

ResponseMeta gives no new answer, deferring entirely to its Question 1 response on the EARN IT Act.

QuestionAsked how a minor is expected to be able to distinguish safe from predatory communication given Meta's control-based safety framing.

ResponseRather than addressing minors' cognitive ability to judge safe vs. predatory contact, Meta answers with a list of default privacy/messaging restrictions applied automatically to under-16 accounts.

QuestionAsked how Meta determines its NCMEC reports are "actionable" and whether current/former prosecutors review them.

ResponseMeta confirms its specialist teams include former FBI investigators and federal child-safety prosecutors, reports all found CSAM to NCMEC's CyberTipline, and cites its 2022 volume of over 26 million NCMEC reports (versus under 6 million for the rest of industry combined).