How the Age Prediction Feature Will Be Introduced

OpenAI is introducing an Age prediction feature to the ChatGPT consumer plans to identify users under 18 and apply appropriate protective measures. This is part of the 'Teen Safety Blueprint' aimed at ensuring the safety and well-being of teens.

The age prediction model comprehensively analyzes behavioral and account-level signals, including account creation period, activity time patterns, usage patterns, and the age users have entered themselves. If an adult is mistakenly classified as a teen, they can immediately restore adult account privileges through selfie verification via Persona, a secure identity verification service.

If a user is determined to be under 18, exposure to the following sensitive content is automatically restricted:

• Violent or graphic content
• Viral challenges that encourage dangerous behavior
• Sexual, romantic, or violent role-play
• Depictions of self-harm and appearance-related harmful content

Additionally, parents can use Parental controls to set usage time limits, control memory and model training, and set up alerts for unusual signs. The feature is expected to roll out in the European (EU) region within a few weeks, in line with regional regulations.