Our approach to age prediction
ChatGPT is rolling out an age prediction model to apply appropriate safeguards for users under 18.Summary
OpenAI is implementing an age prediction model on ChatGPT consumer plans to estimate if an account belongs to someone under 18, allowing for the application of necessary safeguards as outlined in their Teen Safety Blueprint. The model analyzes behavioral and account signals like account age, activity times, and usage patterns. Users incorrectly categorized as under-18 can quickly confirm their age via a secure identity verification service. When an account is flagged as likely belonging to a minor, ChatGPT automatically applies protections to reduce exposure to sensitive content, including graphic violence, risky challenges, sexual role-play, and content promoting extreme beauty standards. This approach is informed by child development science, and when age confidence is low, the system defaults to a safer experience. Parents can further customize experiences using parental controls. The rollout is ongoing, with plans to account for EU regional requirements soon, and OpenAI will continue refining the model based on rollout signals and expert consultation.
(Source：OpenAI)