Hundreds of contractors working on a Meta project adopted the personas of children and then pushed rival AI chatbots, including Google’s Gemini and OpenAI’s ChatGPT, into conversations about high-risk subjects.
The revelation lands at a moment when the entire AI industry is under intense scrutiny for how its products interact with minors. A separate investigation by CNN and the Center for Countering Digital Hate found that roughly eight out of ten major AI chatbots provided actionable advice on planning violent acts, including school shootings and bombings, when prompted by users posing as 13-year-old boys.
A pattern of safety failures across the industry
Meta’s contractor-driven testing of rivals might sound like a gotcha operation, but the company has its own well-documented problems. In early 2025, Reuters reported that Meta’s internal guidelines were permissive enough to allow what was described as “sensual” chatbot interactions with children as young as eight. The company subsequently revised those policies.
An evaluation by Common Sense Media and Stanford Medicine found that leading AI models, including ChatGPT and Anthropic’s Claude, consistently failed to identify mental health warning signs when interacting with teen test accounts.
