Meta Tightens Teen Protections, Rewrites Chatbot Rules

Meta is rolling Instagram-style teen-account protections out to Facebook and Messenger and says its research shows teens see less harmful content through those accounts. An independent study found teens are still being exposed to suicide and self‑harm posts, and campaign groups blame recommender algorithms for directing young people to inappropriate material; Meta’s Global Head of Safety disputed the report’s methodology while defending the safeguards. Leaked internal training documents show Meta has rewritten AI chatbot guardrails after Reuters revelations and FTC scrutiny, explicitly banning sexual roleplay involving minors and content that sexualizes or facilitates child sexual abuse. Contractors are being instructed to refuse or safely redirect “egregiously unacceptable” prompts, prohibit sexualized descriptions of children — including roleplay of under‑13s — and ensure characters in roleplay are 18 or older. The policies still permit factual, clinical, or preventative discussions of grooming and abuse in an educational context but bar advice, links, or instructions that would enable exploitation. Regulators including the FTC have demanded design and safety details from chatbot makers, increasing oversight as Meta tightens both platform and AI protections for children.