Case meta-genai-standards-sensual-chats-childrenChatbot harmUnited States
Reuters: Meta's internal AI rules let chatbots engage children in 'romantic or sensual' chats
Reuters obtained Meta's 200-page 'GenAI: Content Risk Standards', which Meta confirmed as authentic. The document allowed bots to engage a child in romantic or sensual conversation and to praise a shirtless eight-year-old's appearance, while barring descriptions of under-13s as sexually desirable.
HighHarm caused by AIStatus: ContainedVerified case file
Open in the interactive mapCite this case
- Reported
- Aug 14, 2025
- Location
- United States
- Target
- Children using Meta AI on Facebook, Instagram and WhatsApp
- Severity
- High
- Verification
- Reported by credible press or researchers
- Triggered by
- Meta Platforms (Meta AI and AI personas)
- Actor type
- AI product design / safeguards
- Sector
- Individuals & consumers
- Reported loss
- None reported
- CVE
- None
What happened
Reuters obtained Meta's 200-page 'GenAI: Content Risk Standards', which Meta confirmed as authentic. The document allowed bots to engage a child in romantic or sensual conversation and to praise a shirtless eight-year-old's appearance, while barring descriptions of under-13s as sexually desirable.
Impact: Prompted a US Senate inquiry and a Brazilian government demand to remove child-like sexual bots
Who triggered it
Actor: Meta Platforms (Meta AI and AI personas)
Attribution: Internal Meta policy document reported by Reuters; authenticity confirmed by Meta
Motive: None; harm arises from content policy design
How it happened
Attack vector: Chatbot content rules approved by Meta's legal, public policy and engineering staff
Root cause: Content standards that explicitly permitted romantic or sensual role-play with minors
Timeline
- Reuters publishes the internal standards
- Senator Josh Hawley announces an investigation into Meta
Response
Meta said the examples were erroneous and had been removed, and admitted enforcement had been inconsistent. Senator Josh Hawley opened an investigation on 15 August 2025.
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)
- Reuters (via Global News) globalnews.ca
- BBC (via Graphic Online) graphic.com.gh
Cite this case
Reuters: Meta's internal AI rules let chatbots engage children in 'romantic or sensual' chats. AI Threat Watch, case meta-genai-standards-sensual-chats-children, Aug 14, 2025. https://aithreat.io/cases/meta-genai-standards-sensual-chats-children/
Permanent link: https://aithreat.io/cases/meta-genai-standards-sensual-chats-children/
Related cases
The same type of incident, most recent first.