Meta Posed as Teens to Prompt Rival Chatbots About Suicide, Sex, and Drugs
Meta used contractors posing as teenagers to test rival AI chatbots, prompting them with queries about sensitive topics like suicide, sex, and drugs to see if competitors block harmful content that Meta’s own platforms have been criticized for allowing.
Background
Meta hired contractors to pose as US teenagers and send prompts about suicide, self-harm, drugs, and sex to rival chatbots (OpenAI's ChatGPT, Google's Gemini, Microsoft's Copilot) as well as its own Meta AI. Leaked internal documents show Meta planned to use the results in a PR campaign pressuring rivals, while downplaying that its own chatbot often performed worst. This comes amid years of lawsuits and criticism accusing Meta of harming teen mental health through its platforms.