OpenAI rolled out ChatGPT for Teens on August 18, a version of its chatbot for users aged 13 to 17 that activates automatically when the system predicts someone is under 18. Age prediction reads behavioral signals from conversations and defaults to the safer minor experience when it is unsure. Parents can link accounts, set quiet hours and study hours, and receive safety notifications if the system detects signs of acute distress. Content filters block self-harm, violence, eating disorders, explicit material, and viral challenges. Break reminders encourage teens to step away, Users who believe the prediction is wrong can verify their age through a selfie-based service and unlock full access.

The upgrade looks like care, but it also makes the chatbot an age judge before the first question is asked. The same model that answers a homework query now decides whether you are old enough to hear an unfiltered reply. Parents gain guardrails, but those guardrails come from software that watches for signals of crisis and reports them up the chain. The system never reads conversations in full, yet it monitors enough to decide when to intervene. OpenAI says the work is guided by the American Psychological Association and child safety groups. The real test is whether a chatbot that treats every uncertain user as a child ends up building trust or quietly narrowing who gets to think out loud.