ChatGPT Adds Age Prediction Features, Here’s How It Works

OpenAI has introduced an age prediction feature for ChatGPT consumer plans to identify accounts likely belonging to individuals under 18. This tool is designed to enforce appropriate safeguards for teen accounts while allowing adults to use the service under standard safety regulations.

This initiative aligns with OpenAI's Teen Safety Blueprint and guidelines on interacting with users under 18. The company emphasizes the importance of providing young users with learning and creativity tools while ensuring their wellbeing. Age prediction complements existing safety measures for users who indicate they are under 18 during registration.

The global rollout of this feature has commenced, with a slight delay in the European Union to ensure compliance with regional regulations.

The age prediction model evaluates signals associated with an account, such as account age, usage timings, usage patterns, and declared age. The system's performance during the rollout will inform updates to enhance accuracy. OpenAI acknowledges the limitations of automated systems in achieving complete accuracy.

Users mistakenly categorized as under 18 can verify their age through Persona, a third-party identity verification service, using either a live selfie or government-issued ID, depending on the user's country. This verification can be initiated via account settings. Once confirmed as 18 or older, the additional safety settings are lifted, though there might be a brief delay in reflecting these changes across the account.

Accounts identified as likely belonging to users under 18 are immediately subject to enhanced safety settings that restrict access to sensitive content. This includes material related to graphic violence, gory content, risky viral challenges, inappropriate role play, self-harm, and extreme beauty standards.

These guidelines are informed by research on child development, considering differences in risk perception and emotional regulation during adolescence. If age signals are unclear, the system defaults to a safer user experience.

Teen users retain the ability to engage in learning and creative activities on ChatGPT, although content handling may vary based on topic sensitivity. Parental controls are available to further customize the experience, including setting quiet hours and managing memory or training features. Alerts are provided if the system detects signs of acute distress.

Privacy is integral to the design, with Persona handling age verification and deleting selfies or ID images within seven days. OpenAI does not receive copies of IDs or images; only the date of birth or confirmation of being 18 or older is stored in accordance with the privacy policy.

Users who opt out of age prediction can verify their age, after which age checks are no longer conducted on their account.

Based on reporting by techround.co.uk.