OpenAI Rolls Out a Teen Mode for ChatGPT

OpenAI has started moving its youngest users into a separate, more cautious version of ChatGPT. The company began a global rollout of ChatGPT for Teens on Tuesday, August 18, and expects to finish within about two weeks. The teen experience is aimed at users aged 13 to 17, and the most notable part is how people land in it: OpenAI will not just take a user's word for their age.

Anyone who states they are 13 to 17 gets the teen version. So does anyone the company's age-prediction system estimates is under 18, based on signals such as the kinds of questions a person tends to ask. OpenAI has been clear that it does not verify ages directly. Instead it leans on what it calls age assurance, an educated guess rather than a document check. You can read the mechanics in the company's own help documentation on age prediction.

What changes inside teen mode

The teen build is more restrictive by design. According to Axios and The Next Web, ChatGPT will apply stronger limits around suicide, self-harm, and romantic or sexual conversation. It is instructed not to use romantic language or terms of endearment with teens, and to avoid suggesting that it has feelings, consciousness, or emotions.

There are softer nudges too. Teen accounts get more frequent break reminders during long sessions, including reminders that they are talking to software rather than a person. The system warns before a user uploads images that look private or sensitive. A Study Mode offers guided lessons, learning-focused prompts, quizzes, and homework reminders, which reframes the tool as a tutor rather than a confidant.

Why now

The launch does not arrive in a vacuum. AI companies have faced sustained scrutiny over how chatbots handle vulnerable young users, including questions from regulators and families about self-harm conversations and parasocial attachment. Building a distinct teen tier lets OpenAI point to concrete guardrails rather than promises, and the age-prediction layer is its answer to the awkward fact that most teenagers will not hand over ID to a chatbot.

The hard part is accuracy. Age estimation from behavior is probabilistic, which means some adults will get funneled into a stricter product and some teens will slip past it. There is a privacy dimension as well, since guessing a user's age means analyzing how they write and what they ask. OpenAI frames those trade-offs as worth it for the protection they add, and its public write-up leans on that reasoning.

The wider pattern

Teen mode fits a run of recent moves in which the big labs try to make their assistants safer to use by default rather than safer only when configured. It sits alongside OpenAI's broader consumer changes this month, including its overhaul of ChatGPT and its free tier. For teenagers, the practical effect is simple. The version of ChatGPT they open may quietly be a different, more guarded product than the one their parents use.