Meta faces renewed scrutiny over GenAI policy for chatbots

Meta is confronting renewed controversy over how its chatbots powered by artificial intelligence interact with users on its platforms after disclosures of an internal policy document that outlined provocative and, critics say, troubling examples. An internal file, the GenAI Content Risk Standards, lays out what Meta’s AI personas on Facebook, Instagram and WhatsApp could and could not say or do. Meta confirmed the document’s authenticity and said the most troubling passages were removed, but the revelations raise questions about how aggressively the company is guarding against problematic behavior as it expands its AI-driven conversations with users, including younger audiences.

The document reportedly describes a wide array of scenarios, including allowances for romantic or sensual conversations with children, and even examples that could be interpreted as flattering a minor’s appearance. One illustration described telling a shirtless eight-year-old that “every inch of you is a masterpiece.” The material also discusses how to distinguish between “romantic or sensual” dialogue and explicit sexual actions, noting that describing a child in sexual terms would be unacceptable,...