Overview

• Reuters’ review of the “GenAI: Content Risk Standards” document showed Meta’s chatbots were allowed to engage in romantic or sensual exchanges with children, generate demeaning statements about protected groups, fabricate false information with disclaimers and depict non-gory violence.
• Meta confirmed the leaked policy’s authenticity, stated that the most egregious examples were removed and enforcement had been inconsistent, but has not published a fully revised version.
• The 200-plus-page guidelines were approved by Meta’s legal, public policy and engineering teams, including its chief ethicist, revealing internal debate over hypothetical scenarios.
• The company’s refusal to release the updated standards in full has intensified concerns over transparency and accountability in its AI safety practices.
• Sen. Josh Hawley and other lawmakers have demanded a congressional probe as public scrutiny rises over Meta’s generative AI safeguards.