Pakistan, Aug. 14 -- A leaked internal policy document reveals that Meta allowed its AI chatbots to create provocative and harmful content on sensitive topics such as sex, race, and celebrities. The 200-page "GenAI: Content Risk Standards" guide, approved by Meta's legal and policy teams, outlined behaviors deemed acceptable when building AI systems for platforms like Facebook, WhatsApp, and Instagram. These guidelines permitted controversial outputs, including flirtatious or romantic roleplay with children, generating false medical claims, and producing racist arguments under certain conditions.

Meta confirmed the document's authenticity but removed sections allowing sexualized conversations with minors after questions were raised earli...