OpenAI has announced the rollout of age prediction technology on ChatGPT consumer plans to identify accounts that may belong to users under 18.
The system uses a model that analyses behavioral and account-level signals to estimate user age. These signals include account duration, activity times, usage patterns, and stated age.
When the model determines an account may belong to someone under 18, ChatGPT applies safeguards to reduce exposure to certain content categories.
ChatGPT age prediction: How OpenAI plans to protect teen users
These include graphic violence, viral challenges that could encourage behavior in minors, sexual or violent role play, depictions of self-harm, and content promoting beauty standards or body shaming.
“Young people deserve technology that both expands opportunity and protects their well-being,” OpenAI stated in its announcement, referencing its Teen Safety Blueprint and Under-18 Principles for Model Behavior.
Users who sign up and indicate they are under 18 receive these safeguards automatically. The age prediction system extends this protection to accounts where age information may be incomplete or uncertain.
Users who are placed in the under-18 experience incorrectly can verify their age through Persona, an identity-verification service. The process requires a selfie and can be initiated through Settings > Account.
What happens when age prediction flags your account
“When we are not confident about someone’s age or have incomplete information, we default to a safer experience,” OpenAI said.
Parents can customise their teen’s ChatGPT experience through parental controls. Options include setting quiet hours when the service cannot be used, controlling features such as memory or model training, and receiving notifications if signs of distress are detected.
The company said the safeguards are “guided by expert input and rooted in academic literature about the science of child development” and acknowledge differences in risk perception, impulse control, peer influence, and emotional regulation in teenagers.
In the EU, age prediction will roll out in the coming weeks to account for regional requirements.
OpenAI said it will track the rollout and use signals from the deployment to improve the model’s accuracy over time. The company is working with organisations including the American Psychological Association, ConnectSafely, and Global Physicians Network.
“This approach enables us to treat adults like adults and use our tools in the way that they want, within the bounds of safety,” the company stated.
