OpenAI launches ChatGPT for Teens

OpenAI launched ChatGPT for Teens on August 18, 2026, describing it as an experience designed to help teens learn, with stronger built-in safety protections and additional controls for parents. It applies to existing ChatGPT accounts automatically: anyone whose age OpenAI's system estimates as under 18, or who says they are 13 to 17, is placed into it. OpenAI has previously called that estimating system Age-Prediction. Adults who are not classified as teens see no change.

How the safety rules change

The protections are on by default for everyone in the teen experience and follow the Under-18 Principles in OpenAI's Model Spec. The higher-risk areas the company names are self-harm, violence, eating disorders, dangerous activities, and explicit sexual or graphic content.

The under-18 rules also govern how the model talks. ChatGPT is not supposed to use romantic language, encourage emotional dependence, or imply that it has feelings or consciousness, and romantic or sexualized roleplay is blocked. Around the conversation itself, teens get break reminders, cues identifying ChatGPT as AI, and a caution against uploading private pictures.

How homework help works

OpenAI pitches the product as support for learning outside the classroom, while ChatGPT for Teachers remains the classroom offering. When a teen asks for a quick answer, the product can steer them toward working through the problem instead. Study Mode does this with guiding questions and step-by-step support, and new responsible homework reminders are meant to notice when a teen appears to be shortcutting an assignment and redirect them into Study Mode. Study Hours let a teen or a parent pick times when Study Mode is on by default.

OpenAI says its early evaluation of tools like Study Mode has shown promising gains in student performance, but it has not published those numbers.

What parents can control

A parent who links to a teen's account can set Quiet Hours, manage selected settings, and receive safety notifications in limited high-risk situations. OpenAI says it is adding further notifications related to eating disorders while limiting what gets shared with the parent. What linking does not do, at least in what OpenAI has described so far, is give parents a way to read their teen's chats. The company also has not explained how the linking process works.

What is still unknown

The open questions cluster around the age estimate. OpenAI has not published an error rate for it, has not said how a misclassified adult exits the teen experience, and has not shared bypass rates for the new refusals. Until someone outside the company measures the safety and learning claims, they are OpenAI's descriptions of its own product.