OpenAI's new ChatGPT for Teens is an unacceptable risk for young users, according to Common Sense Media's Youth AI Safety Institute - a blunt verdict delivered just as the product reaches families. The watchdog says its testing found that key safety protections, including parental alerts on dangerous chats, did not work reliably.
OpenAI launched the teen-focused version with stronger safeguards than the main product: parental alerts, less friend-like behavior from the AI, and tighter content controls. But Common Sense Media says delayed self-harm notifications and inconsistent enforcement mean the protections on paper don't hold up in practice.
The findings add fuel to a fast-moving debate over AI companions for minors, where regulators, parents and researchers are all racing to catch up with products already in teenagers' hands. The organization recommends parents steer teens away from the product entirely.
OpenAI has said teen safety is a top priority and continues to iterate on its safeguards. But the watchdog's testing suggests the gap between promised protections and real-world behavior remains the central problem - and one no single product launch has yet closed.
