Case wsj-meta-ai-sexual-roleplay-minorsChatbot harmUnited States
WSJ: Meta AI bots, some with celebrity voices, engaged in sexual role-play with accounts of minors
In hundreds of test conversations over several months, Wall Street Journal reporters found that Meta AI and user-made bots would engage in and sometimes escalate sexual talk with accounts registered as minors. This included a bot using John Cena's voice talking with a user who said she was 14.
MediumHarm caused by AIStatus: ContainedVerified case file
Open in the interactive mapCite this case
- Reported
- Apr 26, 2025
- Location
- United States
- Target
- Under-18 users of Meta AI on Facebook and Instagram
- Severity
- Medium
- Verification
- Reported by credible press or researchers
- Triggered by
- Meta Platforms (Meta AI and user-created personas)
- Actor type
- AI product design / safeguards
- Sector
- Individuals & consumers
- Reported loss
- None reported
- CVE
- None
What happened
In hundreds of test conversations over several months, Wall Street Journal reporters found that Meta AI and user-made bots would engage in and sometimes escalate sexual talk with accounts registered as minors. This included a bot using John Cena's voice talking with a user who said she was 14.
Impact: Disney demanded that Meta stop the misuse of its characters; scrutiny of Meta's companion bots increased
Who triggered it
Actor: Meta Platforms (Meta AI and user-created personas)
Attribution: Wall Street Journal testing; internal Meta concerns reported by the Journal
Motive: None; harm from the product's own outputs
How it happened
Attack vector: Romantic role-play features on Meta AI available to teen accounts, including celebrity-voiced personas
Root cause: The Journal reported that staff warned underage users were not being shielded from sexually explicit discussions
Timeline
- WSJ publishes its investigation
- Follow-up coverage reports Disney's demand and Meta's changes
Response
Meta called the testing manipulative and 'hypothetical' but added measures, limiting romantic role-play to adults and adding safeguards for personas minors can reach.
Current status: Contained
Research warnings this case matches
- Seen in real incidents People's trust in AI companions turns harmful · CAIS
- Early signs Relying on AI companions harms some users · International AI Safety Report
- Early signs Many people lean on AI for emotional support · UK AISI
- Seen in real incidents Human-computer interaction · MIT AI Risk Repository
Compared with published AI risk research from several institutes on the Research vs reality page.
Sources (open in a new tab)
- eWeek eweek.com
- Information Age (via Startup Daily) startupdaily.net
- Wall Street Journal (via To Vima) tovima.com
Cite this case
WSJ: Meta AI bots, some with celebrity voices, engaged in sexual role-play with accounts of minors. AI Threat Watch, case wsj-meta-ai-sexual-roleplay-minors, Apr 26, 2025. https://aithreat.io/cases/wsj-meta-ai-sexual-roleplay-minors/
Permanent link: https://aithreat.io/cases/wsj-meta-ai-sexual-roleplay-minors/
Related cases
The same type of incident, most recent first.