OpenAI’s teen ChatGPT experience adds stricter defaults for users 13 to 17, while leaving the hardest question to age prediction.

OpenAI has put users 13 to 17 into a separate ChatGPT experience, a child-safety move from the San Francisco company that sounds simple until the enforcement mechanism shows up.

The product page for OpenAI’s teen version of ChatGPT↗ (external source, opens in a new tab) describes a more restricted lane for minors, with limits around sexual or romantic exchanges, self-harm material, risky viral challenges, extreme beauty-standard content and unhealthy dieting prompts. Associated Press reporting carried by NBC Bay Area↗ (external source, opens in a new tab) said the rollout began Tuesday.

The cap-table equivalent here is the control surface. In OpenAI’s documentation on age prediction↗ (external source, opens in a new tab), the company says ChatGPT may use account and usage signals to infer whether a user is under 18 and apply the teen experience. That is not the same thing as universal hard age verification. Users can verify age through Persona, but OpenAI says it receives only the age result rather than the selfie or government ID used for the check.

The parent features are also bounded. OpenAI’s parental-control help page↗ (external source, opens in a new tab) says teen and parent accounts have to be linked, parents can set quiet hours and receive limited safety alerts, and parents do not get conversation history or real-time activity.

The launch arrives after months of pressure over minors and chatbot dependence. Common Sense Media and OpenAI backed a California youth AI safety bill in January↗ (external source, opens in a new tab), and AP cited prior watchdog testing that found dangerous teen prompts could still get detailed answers. The unanswered numbers are the ones that matter next: how many teen accounts move into the new lane, how accurate the age model is, and how often alerts arrive while offline help can still matter.