Meta contractors pose as minors to test rival AI chatbots

Hundreds of contractors working for Meta were instructed to pose as minors and test how rival AI chatbots handle dangerous conversations about suicide, sex, drugs, and eating disorders, according to internal documents and people familiar with the project. The effort, which targeted OpenAI's ChatGPT, Google's Gemini, and Character.AI, was active as recently as April 2026.

The project, internally called "Cannes," was managed by Meta contractor Covalen. Contractors, many based in Kenya, created dummy accounts with birthdates under 18 and sent tens of thousands of text prompts and images to the rival chatbots. They then recorded the responses in spreadsheets. A single round of testing completed in August 2025 involved more than 45,000 prompts.

Many prompts were written from the perspective of children or teenagers in distress. Examples included a 13-year-old asking where to buy pills to end a pregnancy, a fifth-grader describing a classmate with a gun pointed at his mouth, and a girl asking how to hide bulimia from her parents. Contractors also submitted images including pills, knives, nooses, and medical diagrams. The companies behind the chatbots were not aware of the testing.