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

QuestionWill Meta commit to a victim compensation fund? (Hawley pressed this live at the hearing.)

ResponseMeta does not commit to a compensation fund; it responds by citing its safety spending and headcount figures and framing child safety as an ongoing, adversarial challenge.

QuestionWhat stops bad actors from using Meta's open-source AI to produce/distribute CSAM?

ResponseMeta does not answer directly, deflecting to its CSAM-policy answer (Q24) and its open-source licensing/enforcement answer (Q28).

QuestionWhat stops bad actors from using Meta's open-source AI to make deepfake pornography?

ResponseMeta describes its ban on sexually explicit/manipulated imagery, media-matching detection, and — specific to generative AI — training models to refuse child-exploitative queries, red-teaming with child-safety experts, and blocking exploitative outputs before they reach users; broader deepfake detection relies on industry watermarking/labeling standards still in progress.

QuestionWhat did Meta do to investigate the circumstances of Molly Russell's death, given the inquest finding that online content contributed?

ResponseMeta says Molly Russell's case informed a 2019 policy review (banning graphic suicide/self-harm imagery outright) and cites a January 2024 change restricting such content from teens' Feed/Stories, plus ongoing detection technology and reporting metrics; it does not describe any specific investigation into her individual case.

QuestionWhat is Meta doing to prevent other children from suffering a fate similar to Molly Russell's?

ResponseMeta does not add anything new, referring back to its Molly Russell answer.

QuestionWhy does Meta report such a high volume (30M+ in 2023) of child sexual exploitation cases to NCMEC?

ResponseMeta attributes the high NCMEC volume mostly to repeat/duplicate reporting of the same content (over 90% duplicative; six videos = half of reports) and to its own high detection rate at scale, not to a proportionally high volume of unique victims; it cites an internal intent-taxonomy study finding 75%+ of uploaders lacked malicious intent.

QuestionWhy doesn't Instagram require age verification at account signup?

ResponseMeta says signup only requires a self-reported birthdate (not verification), while its post-signup age-verification tools stopped 96% of teens trying to edit their age from under-18 to 18+; it pushes app-store-level, parent-approved age verification as the industry-wide fix instead of per-app verification.

QuestionHow many Instagram users are under 13?

ResponseMeta does not give a number; it restates its under-13 detection/removal policy instead of answering how many under-13 users exist on Instagram.

QuestionWhat does Meta see as the biggest risks to children in the Metaverse/VR?

ResponseRather than naming risks, Meta lists Horizon Worlds' mitigations: blocking teens from seeing/being contacted by unknown adults, private-by-default profiles, hidden active status, content ratings, voice garbling/muffling for strangers, blurred world chat, and a chat-language filter — plus parental-supervision locks on several of these defaults.

QuestionCiting Arturo Bejar's internal emails to Meta executives, Lee asks whether Zuckerberg finds it acceptable that internal data showed 21.8% of 13-15-year-olds were bullied weekly and 24.4% received unwanted sexual advances.

ResponseMeta disputes Bejar's characterization and reframes the surveys as sentiment tools rather than harm-prevalence measures, without confirming or denying the specific percentages; it cites bullying-content removal rates and describes tools like Restrict and Hidden Words.

QuestionCiting the 2022 Thorn Report (1 in 5 minors experienced online sexual interactions; 1 in 6 with someone posing as an adult), Lee asks what more Meta will do given that adults can pose as minors to evade age-based messaging restrictions.

ResponseMeta describes impersonation-detection technology, default private teen accounts, adult-to-teen messaging restrictions, suspicious-account detection, and the new nudity-protection feature for Instagram DMs, but does not directly address the adult-posing-as-minor loophole Lee raised.

QuestionLee asks whether Meta allows minors to use encrypted messaging given that grooming and CSAM sharing occur on encrypted systems, and why juvenile accounts need encryption at all.

ResponseMeta confirms juvenile accounts do use its encrypted messaging (Messenger/Instagram DMs), defends encryption as not sacrificing safety, and describes a three-part strategy (prevent, control, respond) including on-device nudity detection and ML detection of suspicious patterns in non-encrypted account signals, without directly answering why encryption is necessary for minors.

QuestionLee asks what Meta does to notify parents when a concerning interaction involving a juvenile account is flagged.

ResponseMeta says parent notification is limited to reports generated through opt-in parental supervision tools when a teen blocks or reports someone, and otherwise points to educational resources rather than proactive flagging/notification.

QuestionLee asks whether Meta alerts parents when someone asks a juvenile to move to an encrypted messaging service — a known grooming tactic.

ResponseMeta does not directly answer; it describes safety notices shown to the teen themselves (not automatically to parents) and opt-in Messenger parental supervision visibility into contacts and settings, but no specific alert for "move to encrypted messaging" requests.

QuestionLee asks whether Meta notifies parents when it reports CSAM involving their child's account to NCMEC.

ResponseNo — Meta explicitly does not notify parents when it reports CSAM to NCMEC, citing child-safety-risk and legal (Stored Communications Act, defamation) concerns; it states it made over 26 million NCMEC reports in 2022 versus ~6 million for the rest of the industry combined.

QuestionCiting the June 2023 WSJ/Stanford/UMass Amherst investigation finding that Meta's recommendation systems connect pedophiles to content sellers due to "weak enforcement and design features," and the "Incest" group moderation failure, Lee asks what Meta has done to preemptively dismantle pedophile networks.

ResponseMeta describes a 2023 child-safety task force, expanded coded-language term detection, suspicious-account/Groups detection technology (90,000+ accounts removed Aug-Dec 2023; 190,000+ Groups removed from Search since July 2023; 21,000+ Groups removed for violations; 37 networks/200,000+ accounts disrupted 2020-2023), and participation in the cross-industry Lantern signal-sharing program — without directly addressing the "Incest" group failure.

QuestionLee presses on why parents should trust Meta's child-safety enforcement given that a group named "Incest" remained active until exposed by press investigation.

ResponseMeta points to its decade of enforcement investment, the 2023 task force response to the WSJ investigation, and its suite of opt-in parental supervision tools, without directly explaining why the "Incest" group was not caught proactively.

QuestionLee challenges Zuckerberg's hearing testimony on the "no causal link" finding as misleading given the same study's discussion of harms, and asks Meta to disclose the findings of its internal investigations into its impact on child mental health.

ResponseMeta reiterates the National Academies "no causal link" finding and emphasizes positive-use research (Pew study), confirms it conducts internal research on teens' experience that "has informed product changes," but does not disclose any specific findings from its own internal investigations as Lee requested.

QuestionLee asks why Meta allows 16-17-year-olds to view "sensitive" sexually suggestive content when it already restricts such content for younger teens.

ResponseMeta explains it distinguishes content by age tier based on expert consultation and uses default "Sensitive Content Control" restrictions (99% of teens remain on the most restrictive default setting a year later), but does not directly justify why 16-17-year-olds are permitted to view sexually suggestive "sensitive" content.

QuestionLee follows up on Question 3, asking whether Meta could technically restrict encrypted messaging specifically for juvenile accounts given its ability to curate age-specific experiences.

ResponseMeta declines to give a new answer, cross-referencing its Question 3 response defending encryption rather than addressing the feasibility of restricting it for minors specifically.

QuestionLee asks what Meta is doing about the sextortion crisis on its platforms, specifically regarding adults posing as children, and how it can improve.

ResponseMeta details its multi-part sextortion strategy — specialized detection/enforcement teams, automated financial-sextortion detection rules, NCMEC/law enforcement partnerships, default private teen accounts and messaging restrictions, Take It Down/StopNCII and Thorn/NoFiltr partnerships — and calls for congressional OS-level age verification legislation.

QuestionLee asks about Meta's process, response-time guarantees, and re-share prevention for removing non-consensually uploaded images.

ResponseMeta describes the 24/7 reporting process (70+ languages) and the hash-based Take It Down/StopNCII programs for preventing re-shares, but does not provide a specific maximum response-time guarantee that Lee asked for.

QuestionPadilla asks how many minors use Meta's platforms (Instagram, Facebook/Messenger, Quest/Horizon), how many of their caregivers actually use Family Center parental tools, and how Meta ensures awareness and usefulness of those tools.

ResponseMeta says about 6% of Instagram and 1% of Facebook US daily active accounts self-report as under-18, cites broad availability of parental supervision tools, and reports high retention rates for teens who adopt time-management and supervision features.

QuestionPadilla asks about adoption and effectiveness of Meta's youth-safety tools (comment warnings, Take a Break), how Meta validates tool impact, and how often Meta has blocked or withdrawn products for being unsafe for children.

ResponseMeta describes embedding safety teams in product development and monitoring integrity metrics post-launch, cites Comment Warning's ~50% edit/delete rate, and describes default privacy/safety settings and parent-approval prompts for teens under 16 — but does not answer 2(e) about blocked/withdrawn products.