OpenAI Age Prediction for ChatGPT

OpenAI has introduced age prediction for ChatGPT consumer plans to identify accounts likely belonging to users under 18. This system allows OpenAI to apply specific safety safeguards and experience modifications tailored to teenagers, while maintaining a standard experience for adults.

Age Prediction Mechanism and Signals

ChatGPT utilizes a dedicated age prediction model to estimate whether an account holder is under 18. The model analyzes a combination of account-level and behavioral signals to make this determination. These signals include:

• Account age (how long the account has existed)
• Typical activity times of day
• Usage patterns over time
• The user's stated age

When the model lacks confidence in the age estimation or possesses incomplete information, it defaults to a safer experience.

Teen Safety Safeguards and Content Restrictions

When an account is predicted to belong to a user under 18, ChatGPT automatically applies additional protections to reduce exposure to sensitive or harmful content. These restrictions are based on academic literature regarding child development, specifically addressing differences in risk perception, impulse control, and emotional regulation in teenagers.

Specific content restrictions include the prevention of:

• Graphic violence or gory content
• Viral challenges that encourage risky or harmful behavior
• Sexual, romantic, or violent role play
• Depictions of self-harm
• Depictions of extreme beauty standards, unhealthy dieting, or body shaming

Identity Verification and Parental Controls

Users who are incorrectly identified as under 18 can restore full access by verifying their age via a selfie through Persona, a secure identity-verification service. This verification process can be initiated in Settings > Account.

Beyond the system-predicted safeguards, parents have access to customizable parental controls. These features include:

• Setting "quiet hours" to restrict ChatGPT usage
• Controlling model training and memory features
• Receiving notifications if the system detects signs of acute distress

Regional Rollout and Future Development

Age prediction is currently rolling out, with a specific deployment for the European Union scheduled for the coming weeks to meet regional requirements. OpenAI is collaborating with experts from the Global Physicians Network, ConnectSafely, and the American Psychological Association to refine these protections and address attempts to bypass safeguards.