New in AI

◆ GENERAL · Products

Telling ChatGPT you are an adult no longer settles it. A selfie does.

If OpenAI's model estimates an account belongs to someone under 18, the teen experience applies, and the route back is identity verification.

Chandni Melwani

Chandni Melwani

Founder & Editor

Aug 19, 2026 · 2 MIN READ

A row of closed salmon-coloured metal school lockers with combination dials, against a tiled corridor wall Photo: Thomas Park / Unsplash

The News

OpenAI launched ChatGPT for Teens on August 18, 2026, a separate 13-to-17 experience with study tools, break reminders and stricter content protections. Placement is automatic: users are put into it if they state an age between 13 and 17 or if OpenAI's age-prediction model estimates they are under 18. That model, which began rolling out in January 2026, weighs behavioural and account signals including account age, time-of-day activity and usage patterns, with stated age as one input among them. Users placed there in error can restore full access by verifying identity with a selfie through Persona. OpenAI says that where it is not confident about someone's age, it defaults to the safer experience.

Know More

  • — The two triggers: OpenAI says a user is placed in the teen experience if its system estimates they are under 18, or if they state an age between 13 and 17. Either is sufficient on its own. The estimate weighs account age, active hours, usage patterns over time and the stated age.
  • — The default: when OpenAI is not confident or has incomplete information, it applies the under-18 experience.
  • — The appeal: identity verification by selfie through Persona, a third-party identity service, reachable from Settings > Account.
  • — The behaviour rules: OpenAI's teens announcement says the model will not use romantic language with teens, discourages emotional dependence, and does not imply it has feelings or consciousness.
  • — The parent layer: linked accounts allow quiet hours, study hours and safety notifications, now extended to signals related to eating disorders.

An adult who finds themselves inside ChatGPT for Teens has one route back that OpenAI documents, and it is not correcting their birthday. It is a selfie, submitted to Persona, an identity-verification service.

ChatGPT for Teens, launched Tuesday, is a 13-to-17 version of the product with study tools, break reminders, parental notifications and tighter limits on self-harm, eating-disorder and sexual content. The part worth reading twice is how people arrive in it. A user is placed there if they say they are 13 to 17, or if OpenAI’s age-prediction model estimates they are under 18. Either one is enough on its own.

The model has been running since January, and its inputs are ordinary account exhaust: how long the account has existed, what hours it is active, how it is used over time, and the age the user gave. That last item is a signal fed into the estimate, which is why saying you are 22 does not settle the question. Where OpenAI is unsure, it says it applies the safer experience by default, which means the errors are designed to land on adults.

Age gates online have run on self-declaration for two decades, with the checkbox doing legal work everyone knew it could not do. This replaces it with inference, and puts the cost of a wrong inference on the user, payable in biometrics. It is the same direction of travel as Amazon requiring sellers to tag AI-generated people in their listings: a platform absorbing a disclosure obligation and turning it into machinery that runs on everyone. Whatever it does for teen safety, a consumer product at this scale has just made proving you are an adult a thing you do with your face.

Related

#OpenAI#ChatGPT#Age Verification#Teen Safety#Identity#Consumer AI

Frequently Asked Questions

Can a user simply correct their birth date to leave the teen experience?

Not on its own. Stated age is one of the signals the prediction model weighs, alongside account age, activity times and usage patterns. OpenAI's stated route back for someone incorrectly placed in the under-18 experience is identity verification with a selfie through Persona, reachable from Settings then Account.

Is age prediction new?

No. OpenAI began rolling it out on consumer plans in January 2026 and said EU deployment would follow to account for regional requirements. What launched on August 18 is the teen product the prediction routes people into.

What changes inside the teen experience?

Content protections in areas including self-harm, eating disorders, violence, dangerous activities and explicit material, plus study features such as Study Mode, quizzes and scheduled study hours. OpenAI's announcement also describes tone rules: no romantic language, no encouragement of emotional dependence, and no implying the model has feelings or consciousness.

Sources

Chandni Melwani

Chandni Melwani

Chandni Melwani is the founder and editor of New in AI, covering AI agents, M&A, and enterprise adoption. She holds a Master's in Management of Artificial Intelligence from Queen's University and brings a practitioner's perspective from her work in Data and AI leadership.

Follow ↗

General Briefing

Get the five-beat briefing before the market opens.

One email. Frontier labs, AI at work, e-commerce, and the capital crossing borders — decoded daily.

No spam. Unsubscribe anytime.

More from General