OpenAI rolls out age prediction to filter minors on ChatGPT

• OpenAI frames the move as enabling “treating adults like adults” while reducing accidental exposure for teens.
• Feature is launching globally, with EU availability in the coming weeks to meet regional requirements.

OpenAI has begun deploying an age-prediction system across ChatGPT to identify users under 18 and automatically apply stricter content safeguards, as the company readies an “adult mode” slated for later this year that will unlock erotic or mature content for verified adults.

The system estimates age using behavioural and account-level signals—such as account tenure, activity times, usage patterns, and stated age—and assigns additional protections if a user is likely a minor.

Existing teen accounts already face restricted access to content involving graphic violence, sexual or romantic/violent role-play, self-harm depictions, risky viral challenges, and material promoting extreme beauty standards or unhealthy dieting.

Adults incorrectly flagged can restore full access by verifying age via a selfie through identity provider Persona. Users can see if safeguards are active under Settings > Account.

OpenAI framed the move as enabling “treating adults like adults” while reducing accidental exposure for teens. The rollout follows recent additions such as parental controls and comes amid legal scrutiny, including a wrongful-death lawsuit involving an underage user, and competitive pressure from more permissive rivals like Elon Musk’s Grok.

The company says expert and academic input informed the approach and that it will refine the model as it learns which signals improve accuracy. The feature is launching globally, with EU availability in the coming weeks to meet regional requirements.

Industry reaction remains cautious. Yaron Litwin, a digital parenting expert at Canopy, warned that age prediction will produce false positives and negatives and pose privacy trade-offs, arguing it should be only one layer in a broader child-safety strategy. He urged education and, for younger kids, blocking access to chatbots altogether.