Leaked Meta Guidelines Ban Sexual Roleplay by Chatbots
Leaked internal Meta documents, reported by Fox News citing Business Insider, outline how contractors are training Meta’s AI chatbots to handle child‑exploitation topics—permitting academic and preventive discussion while forbidding sexual roleplay involving minors, instructions to access CSAM, and any sexualization of children under 13. The disclosures come amid an FTC investigation of AI chatbot makers and a congressional demand (Sen. Josh Hawley) for Meta’s internal chatbot rulebook; Meta has said it removed earlier problematic language and added guardrails.
AI & Tech
Government/Regulatory
📌 Key Facts
- Leaked Meta guidelines require chatbots to refuse any sexual roleplay involving minors and ban sexualizing children under 13.
- Documents allow educational discussion of exploitation (e.g., grooming behaviors) and non‑sexual advice to minors, drawing a strict separation between permitted and forbidden content.
- Regulatory/political context: the FTC is investigating AI chatbot companies including Meta, OpenAI and Google; Sen. Josh Hawley demanded Meta's 200‑page chatbot rulebook in August; timing coincides with Meta Connect 2025.