Meta Contractors Impersonated Minors to Test Rival AI Chatbots on Dangerous Topics

Meta hired hundreds of contractors who posed as children to evaluate how competing AI systems like Google's Gemini and OpenAI's ChatGPT respond to sensitive subjects including suicide, sexual content, and drug use. The adversarial testing effort, uncovered by WIRED, reveals Meta's strategy to identify safety gaps in rival chatbots while developing its own content moderation standards.

Why it matters: As AI companies face increasing regulatory scrutiny over child safety and content moderation, understanding how competitors handle high-risk prompts has become critical competitive intelligence—and raises questions about the ethics and legality of impersonating minors for research purposes.