OpenAI moved ChatGPT for Teens out of a filter applied to accounts and into a named, global product on Tuesday, 18 August 2026. The rollout reaches eligible accounts on both free and paid personal ChatGPT plans, and OpenAI expects it to finish within about two weeks. The European Union rollout follows in the coming weeks.
A Routing Rule, Not a Toggle
Placement into the teen product runs on a rule OpenAI states plainly in its launch announcement: “If our system estimates someone is under 18 or they state their age is between 13 and 17, they are automatically placed into ChatGPT for Teens.” That sentence describes two separate entrances, not one: an account holder who self-declares an age of 13 to 17, and an account the system estimates to be under 18. Neither entrance requires the user to do anything beyond existing inside the product; the stated age or the estimate does the placing.
For signals that don’t resolve cleanly, OpenAI states a second rule: “When we are not confident about someone’s age or have incomplete information, we default to a safer experience.” Read together, the two rules mean uncertainty resolves toward the teen experience, not away from it. inimino covered the behavioural signals behind the estimate in March, and OpenAI lists them in its Help Centre entry on age prediction; what shipped on 18 August is the product built on top of it, not the classifier itself.
How does ChatGPT decide someone is a teen?
ChatGPT places an account in the teen experience through two entrances: a user who states their age as 13 to 17, or OpenAI’s system estimating the account holder is under 18. For borderline cases, OpenAI’s stated rule is: “When we are not confident about someone’s age or have incomplete information, we default to a safer experience.” OpenAI published no accuracy figure alongside the 18 August launch.
Leaving Costs Something, Staying Doesn’t
Nobody needs to verify their age to remain inside ChatGPT for Teens; verification exists only as the way out. An adult who believes the system misfiled them goes to Settings, then Account, then Verify age, and confirms through Persona, a third-party vendor. The correction is opt-in: placement happens automatically in both directions, but reversing it is a task the user has to start and finish with an outside company.
Inside the teen product, ChatGPT restricts:
- Graphic violence or gore
- Viral challenges that could push risky or harmful behavior
- Sexual, romantic, or violent roleplay
- Content promoting extreme beauty standards, unhealthy dieting, or body shaming
ChatGPT is also barred from romantic language and terms of endearment with teens. The product adds more frequent break reminders, including reminders that the user is talking to an AI, and warnings before uploading potentially private or sensitive images.
Also Shipped on 18 August
The same release also shipped a set of learning and parental-control features. Study Mode guides a user through a problem with questions and step-by-step explanations rather than a direct answer, paired with Study Hours, a Responsible Homework Reminder, quizzes, and learning visualisations. Parental controls come with Quiet Hours, toggles for memory and model training, and safety notifications tied to signs of acute distress.
What This Means for Measuring ChatGPT as a Channel
With no accuracy figure published, the share of adult accounts sitting inside the teen product isn’t observable from outside the company. That’s a second change to what a default account gets inside ten days: OpenAI’s earlier change to what default accounts get swapped the model behind Free and Go chats, and this one changes which restrictions apply to an account before a person opts into anything. Anyone reading ChatGPT outputs as a stand-in for a typical user should treat both defaults as live variables, not settled ones, when the account behind a given output was never verified.