Overview
- OpenAI has begun a limited, region-specific rollout of an age-prediction system that estimates whether an account likely belongs to someone under 18.
- When flagged as a likely minor, ChatGPT restricts exposure to sensitive areas including graphic violence, sexual or violent roleplay, self-harm depictions, dangerous viral challenges, and extreme diet or beauty content.
- The model draws on indicators such as account age, typical hours of activity, usage patterns over time, declared age, and language cues rather than image analysis.
- Adults incorrectly classified as minors can verify through Persona with a selfie or government ID to restore adult settings, with OpenAI stating Persona deletes verification images within seven days.
- OpenAI cites guidance from child development experts and positions the rollout as part of safety efforts during increased regulatory scrutiny and litigation over alleged harms to minors, with EU deployment planned in the coming weeks.