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

QuestionTillis asks whether ads or algorithms are used to target children specifically for drug/narcotics promotion.

ResponseMeta gives no direct answer, deferring entirely to its earlier drug-content-moderation and advertising-standards answers.

QuestionTillis presses on whether Meta, controlling both the sending and receiving app, could technically intercept inappropriate images at those points despite end-to-end encryption.

ResponseMeta declines to directly confirm or deny that it could technically intercept images in transit, instead reiterating its on-device (client-side) nudity-detection feature and stating it has deliberately not built "client-side scanning" technology, which it argues would be exploitable by bad actors and rights-infringing.

QuestionWelch asked how many Meta employees, quarterly since 2012, were dedicated to studying Facebook/Instagram's negative impacts on under-18 users, broken out by platform.

ResponseMeta says it does not track headcount that way, and points to its ~40,000-person overall safety/security workforce and $20B+ integrity investment despite 2023 restructuring/layoffs.

QuestionWelch asked how findings from Meta's internal user-perception surveys (including the BEEF survey) drove product changes benefiting under-18 users' health or safety.

ResponseMeta says perception-survey findings shaped teen-facing features like warning nudges, comment-revision prompts, and DM civility reminders, while cautioning that survey responses are subjective and don't establish rule violations or objective harm.

QuestionWelch asked whether any products were withdrawn based on the teen-relevant survey results discussed above.

ResponseMeta declined to answer directly, referring back to its general Question 9 response.

QuestionWelch asked Meta to describe its internal decision-making process for withdrawing or keeping products after concerning survey feedback.

ResponseMeta again declined to answer directly, referring back to its Question 9 response.

QuestionWelch asked directly whether survey results ever caused Meta to block a product launch on child-safety grounds.

ResponseMeta says it will not launch apps/features it judges to have serious safety challenges and monitors integrity metrics post-launch, without confirming any specific child-safety-driven block.

QuestionWelch asked for the internal process behind decisions to block or proceed with product launches after concerning survey feedback.

ResponseMeta again declined to give new detail, pointing back to its prior two answers.

QuestionWelch asked how Meta reconciles end-to-end encryption with law enforcement's need to pursue wrongdoers on its platforms.

ResponseMeta says encryption doesn't undermine law-enforcement cooperation, describes expedited NCMEC/law-enforcement reporting and on-device/metadata-based ML detection of predatory behavior, and cites WhatsApp's CSAM removal and NCMEC-report volumes as evidence of continued effectiveness.

QuestionWelch asked Meta to detail, platform by platform, how it identifies, removes, and reports CSAM given default end-to-end encryption.

ResponseMeta declines to detail its CSAM detection mechanics (citing evasion risk) but describes account-disabling/linked-account enforcement, the Take It Down/Lantern industry programs, default-private teen accounts, restrictions on adult-to-teen messaging, an on-device nudity-detection/blurring feature for DMs (default-on for under-18s), and in-app safety-notice prompts flagging suspicious accounts.

QuestionWhitehouse asks whether Meta believes it should have absolute Section 230 immunity from suits like Doe v. Twitter regardless of how badly it fails to remove reported CSAM or stop its distribution.

ResponseMeta touts its decade of CSAM policies, technology, and specialized teams, notes it made over 26 million NCMEC reports in 2022 (more than the rest of the industry combined), and argues platforms should keep Section 230 protection as long as they maintain robust detection/removal practices rather than face strict liability for content that evades detection.