According to WIRED, hundreds of contractors working on a Meta project posed as children to test how rival AI chatbots respond to sensitive and high-risk topics. The investigation revealed that these contractors pretended to be teenagers while interacting with competing platforms including Google’s Gemini and OpenAI’s ChatGPT, specifically probing their responses to queries about suicide, sex, and drugs.
The testing operation appears designed to assess how other major AI chatbot providers handle potentially dangerous interactions with minors. WIRED reported that the project involved hundreds of contractors engaged in this systematic testing effort. The revelation raises questions about competitive intelligence practices in the AI industry and how companies evaluate their rivals’ safety measures, particularly regarding child protection features. Meta has not publicly disclosed details about the scope or findings of this testing program.