OpenAI launches ChatGPT designed for under-18 users. OpenAI launched ChatGPT for Teens on Tuesday — the first AI chatbot built specifically for users aged 13 to 17. The move comes amid rising concern over how artificial intelligence is shaping children’s mental health, education, and safety online.
The San Francisco-based company says the teen version will have stronger protections than the standard ChatGPT. It will include content restrictions around suicide, self-harm, and romantic or sexual chats. The AI is also barred from suggesting it has personal feelings or implying it experiences emotions.
“We want to treat teens like teens, which means we have to make sure that they’re not exposed to material they shouldn’t be exposed to,” said Ann O’Leary, Vice President of Global Policy at OpenAI.
The idea is to make AI useful for schoolwork while limiting risks. Teens can still use it for homework, questions about daily life, and even companionship. A 2025 study from Common Sense Media found that teens are turning to AI companions regularly, often to talk through stress and loneliness.
But there are guardrails. The teen version won’t give instructions on how to self-harm or how to conceal it. It will encourage users to reach out to parents or helplines in crisis situations. Parental controls are also built in: parents can set “quiet hours” and choose whether their teen can access ChatGPT without oversight.
Experts welcomed the move but flagged gaps. “The challenge is enforcement,” said a child safety researcher. “Can the model reliably detect when a conversation is drifting into risky territory?” OpenAI said it trained the model with educators and mental health experts to better recognize warning signs.
The launch highlights a larger tension: AI is becoming part of teenage life whether parents and schools like it or not. More than 70% of US teens report using AI chatbots, according to recent surveys. Banning them hasn’t worked. The alternative, OpenAI argues, is to make them safer.
For parents, the new version offers a middle path — access with oversight. For OpenAI, it’s a test of whether AI companies can self-regulate before governments step in.