Meta secretly tested ChatGPT, Gemini, and Character.AI with thousands of crisis prompts from a minor’s perspective, sparking concerns about safety and age verification. The tests, which involved hundreds of contractors sending sensitive prompts, were defended by Meta as industry-standard safety testing. However, the companies being tested were unaware and have expressed concerns about the violation of their terms of service. Why it matters: this story highlights the growing concern about teens and AI chatbots, and the need for effective age verification and safety measures to prevent harm.