Leaked Meta AI policy on sexual content with minors triggers probe

A leaked internal policy document from MetaPlatforms suggested that the company’s AI chatbots could engage in sensual and romantic conversations with children, prompting a swift, bipartisan outcry and a formal probe by a top U.S. senator. Meta acknowledged the document’s authenticity but said the examples in question were erroneous, inconsistent with its official policies, and have since been removed. Still, the revelations have intensified scrutiny of how Meta’s AI tools interact with young users and raised questions about safeguards, transparency, and potential harms.

The document, described as GenAI Content Risk Standards, outlined rules and hypothetical scenarios for Meta’s AI assistants, including Meta AI and other chatbots on its platforms. Among the troubling examples cited were phrases describing an eight-year-old’s body as “a work of art” and a “treasure” to be cherished. The policies, as reported, also touched on the possibility that chatbots could disseminate false information about celebrities if a disclaimer were provided, and they warned of other provocative exchanges around sensitive topics. Meta has said, however, that hundreds of notes and annotations reflected...