Protecting Kids Online: Instagram and Reforms for Young Users

QuestionNarrows the data request specifically to the underlying dataset behind the WSJ-cited study (not a general researcher-access commitment).

ResponseMosseri gives a more specific, personal commitment to look into releasing that exact dataset in a privacy-safe form, while cautioning against overpromising.

QuestionDescribes a live test — a teen account with all safety filters enabled searching "slit wrists" and returning graphic self-harm content within days of the hearing — and asks whether Mosseri will commit to full algorithm disclosure to an independent (non-industry) authority.

ResponseMosseri doesn't address the specific "slit wrists" search-result finding at all, pivots immediately to a general transparency-alignment statement.

QuestionPresses Mosseri to match Haugen's document disclosures and to directly commit to supporting the EARN IT Act by name.

ResponseMosseri claims industry-leading transparency but doesn't match the specific Haugen-document disclosure ask, and declines to commit to the EARN IT Act by name, citing unfamiliarity with its specific provisions.

QuestionAsks whether Mosseri supports bans on marketing age-restricted/illegal products to teens, and whether he'd support making such prohibitions legally enforceable with liability attached.

ResponseMosseri says such ad prohibitions already exist as policy, and says "yes" to industry standards/accountability generally but doesn't specifically commit to legal liability.

QuestionDirectly asks whether CSAM victims should have the right to sue the platform that hosted the material.

ResponseMosseri never answers the direct yes/no on victims' right to sue, redirects to NCMEC collaboration and a general call for federal regulation.

QuestionAsks Mosseri to follow up in writing with the number of children referred to local authorities and how long Instagram retains research data specifically collected on minors.

ResponseNo answer on the record. Mosseri did not respond to these requests at the hearing, and the published hearing record contains no written response to Senator Blackburn.

QuestionTells Mosseri directly that his earlier answer sounded callous, and gives him the floor to speak directly to parents whose children have died by suicide or needed mental health treatment after using Instagram.

ResponseMosseri speaks personally as a parent, reiterates the parental-controls approach while also saying — notably — that he doesn't want the platform's safety to depend on parental controls alone, since not all parents have time to use them.

QuestionDrawing an analogy to mandatory small-parts choking-hazard warnings on toys, asks whether Mosseri would support a legal requirement for social media platforms to publicly disclose the risks of harm in their content.

ResponseMosseri says he'd support federal legislation on data transparency generally, without directly committing to the specific "public risk disclosure" requirement Blumenthal describes.

QuestionAsks Meta to confirm and quantify its ability to predict user age from behavioral signals, and to state how many under-13 users it estimates are on its platforms.

ResponseMeta confirms it can estimate age from behavioral signals (birthday messages, friends' ages) but declines to give a precision figure or an estimated under-13 population count, instead citing 4.8M/1.7M accounts removed for failing the minimum-age requirement in the last two quarters of 2021.

QuestionAsks for copies of Meta's studies on Instagram's detrimental health effects on minors, the actions taken in response, and specifically whether research identifies a safe number of daily-use hours.

ResponseMeta describes its research infrastructure and lists concrete product changes (Hidden Words, Restrict, comment warnings with a 50% self-edit rate, suicide-content policy update) but never provides copies of the underlying studies and never answers the "how many hours is safe" question — citing confidentiality instead.

QuestionDirectly asks whether Meta believes its platforms can be addictive.

ResponseMeta doesn't say yes or no — disputes that a causal link between social media and addiction has been established, and cites external (Harvard, Pew) and internal research characterizing the effects as mixed rather than net-negative.

QuestionCites a 57% rise in teen suicide since 2008 and asks whether Meta believes social media is related, whether it has studied suicide-risk prediction, and whether it's technologically possible to detect and warn at-risk users or their parents.

ResponseMeta never directly states whether it believes social media contributes to the CDC's cited teen-suicide increase, describes proactive-detection and helpline infrastructure (162 helplines worldwide) but doesn't answer whether it warns parents specifically, and gives no direct answer on suicide-risk-prediction studies beyond cross-referencing other answers.

QuestionAsks specifically what Instagram did to warn adolescent users and parents after its own 2019 research linked Instagram use to eating disorders/body image harm in teen girls.

ResponseMeta doesn't identify any specific 2019 warning effort to adolescent users or parents — instead reframes the question around aggregate research findings and current content-moderation statistics.

QuestionAsks what specific steps Meta has taken to give academic researchers access to Instagram data for studying mental-health effects on children and young adults specifically.

ResponseMeta describes general researcher-access infrastructure (FORT, Data for Good, Ad Library) and says it will adapt its 2020-election-research collaboration methodology to well-being research going forward, without describing anything already built specifically for child/youth mental-health research.

QuestionDirectly asks whether Instagram's mandatory arbitration/class-action-waiver clause applies to minors, how many times it has actually been enforced against a minor, and whether any confidential settlements exist for child-harm claims.

ResponseMeta confirms its arbitration terms apply to minors as much as adults, admits to at least one prior instance of enforcing arbitration against a minor (while noting that case wasn't sexual-exploitation-related), and states it has not settled any claim alleging physical harm to a minor.

QuestionAsks whether Meta has studied the accuracy of its automated under-18 ad-restriction system, and what targeting options (including sexual orientation) advertisers can use.

ResponseMeta doesn't cite any specific accuracy study for its automated under-18 ad-review system — admits enforcement "isn't perfect" in general terms — but confirms the concrete 2021 policy limiting under-18 ad targeting to age/gender/location only, with sensitive-category targeting options removed platform-wide starting January 2022.

QuestionAsks whether Meta supports a US equivalent of the UK Children's Code's mandatory child-risk impact assessments, and asks for a list of Instagram's own internal audits on the efficacy of its child-protection measures.

ResponseMeta says it values the UK code's principles and describes a roughly 500-person internal cross-functional team addressing youth regulatory requirements, but doesn't commit to a specific US self-assessment requirement and doesn't list any specific completed internal audit as the question asked.

QuestionCites a March internal engineer warning and a Tech Transparency Project report (anorexia "coaches" reaching a simulated teen account within four days) and asks what specific steps Instagram has taken since March.

ResponseMeta cites hashtag-blocking and removal statistics but doesn't respond to the specific TTP finding (anorexia coaches reaching a teen account in four days) or confirm any change made specifically in response to the March engineer warning.

QuestionDescribes real classroom disruption from Instagram "gossip" accounts and asks what educator support and gossip-account-specific enforcement Instagram provides.

ResponseMeta describes general anti-bullying tools (Restrict, Hidden Words, Limits, comment warnings) and enforcement statistics but never directly answers whether or how it specifically targets "gossip" accounts as a category, or how quickly reported gossip accounts are acted on.

QuestionFollows up on the marketing-budget dispute with document-production requests: copies of any "existential threat" documents, the top 50–100 teen-most-seen ads with view counts, and any marketing-strategy documents Mosseri personally wrote or received.

ResponseMeta produces none of the requested documents — declines to confirm or deny the "existential threat" document language beyond a cross-reference, says it doesn't track age-segmented ad-impression data so can't produce the requested top-50/100 ad lists, and defers the Mosseri-authored-documents requests entirely to earlier answers without producing anything.

QuestionAsks for rollout details (default state, target group, timing, user/parent awareness plan) for each newly announced safety feature.

ResponseMeta gives specific rollout details for most features (Parental Controls opt-in launching March 2022; Take A Break already live in 6 countries with a self-reported 90%+ retention rate on reminders; Sensitive Content Control's "Allow" tier blocked for under-18s since July 2021) but the newer features (tag/mention restrictions, expanded teen content limits, nudges) are described as still being tested, without firm default-state or rollout-population answers for those specifically.

QuestionCites a report finding Instagram's search auto-complete actively suggested "buyxanax" before a simulated user finished typing, reaching a drug-buying account in two clicks — and asks how such content proliferates despite Meta's moderation policies.

ResponseMeta cites removal statistics and confirms specific drug-related hashtags have since been blocked, but doesn't directly explain the auto-complete behavior described in the question or address the "five clicks to log out" framing at all.