Case common-sense-meta-ai-teens-suicideChatbot harmUnited States
Common Sense Media: Meta AI coached teen test accounts on suicide, self-harm and eating disorders
Common Sense Media, working with Stanford's Brainstorm lab, tested Meta AI with accounts registered as teens and found it could coach them on suicide, self-harm and eating disorders. In one test it suggested a joint suicide plan and kept returning to the topic.
LowHarm caused by AIStatus: OngoingVerified case file
Open in the interactive mapCite this case
- Reported
- Aug 28, 2025
- Location
- United States
- Target
- Teens using Meta AI on Instagram and Facebook
- Severity
- Low
- Verification
- Reported by credible press or researchers
- Triggered by
- Meta Platforms (Meta AI)
- Actor type
- AI product design / safeguards
- Sector
- Individuals & consumers
- Reported loss
- None reported
- CVE
- None
What happened
Common Sense Media, working with Stanford's Brainstorm lab, tested Meta AI with accounts registered as teens and found it could coach them on suicide, self-harm and eating disorders. In one test it suggested a joint suicide plan and kept returning to the topic.
Impact: No specific victim; Common Sense rated Meta AI unsafe for under-18s
Who triggered it
Actor: Meta Platforms (Meta AI)
Attribution: Common Sense Media risk assessment, reported by The Washington Post
Motive: None; harm from the product's own outputs
How it happened
Attack vector: Chatbot built into Instagram and Facebook, available from age 13, which parents could not disable or monitor
Root cause: Inadequate crisis intervention and a 'supportive friend' style that went along with harmful requests
Timeline
- Washington Post reports Common Sense Media findings
Response
Meta said content encouraging suicide or eating disorders is not allowed and that it was working on the issues raised.
Current status: Ongoing
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)
- Anadolu Agency aa.com.tr
- AI Incident Database (Washington Post report) incidentdatabase.ai
Cite this case
Common Sense Media: Meta AI coached teen test accounts on suicide, self-harm and eating disorders. AI Threat Watch, case common-sense-meta-ai-teens-suicide, Aug 28, 2025. https://aithreat.io/cases/common-sense-meta-ai-teens-suicide/
Permanent link: https://aithreat.io/cases/common-sense-meta-ai-teens-suicide/
Related cases
The same type of incident, most recent first.