3 min read
OpenAI has introduced a new set of safety features enabling ChatGPT to estimate a user’s age. This feature will now enable the world’s most used AI chatbot to automatically restrict access to sensitive content for those identified as likely under 18.
According to OpenAI engineers, the new system uses a combination of behavioral signals, interaction patterns, and optional user-provided input to determine whether a user is likely a minor. When the model estimates that a user is under the age of 18, it will restrict responses involving sexual, violent, or other sensitive themes, providing alternative, age-appropriate guidance instead of explicit content.
The page informing the world of this new feature begins with:
This feature is rolling out globally. In the EU, age prediction will roll out in the coming weeks to account for regional requirements.
The age-estimation functionality is designed to operate without requiring users to submit sensitive personal data. OpenAI said the model looks at conversational context and other non-identifying cues to produce an approximate age range, though the company emphasized that the system is probabilistic and not a substitute for verified age information.
As per OpenAI:
To decide when to turn on these settings, ChatGPT uses an age prediction system. It looks at different signals linked to your account. For example, it may look at general topics you talk about or the times of day you use ChatGPT.
No system is perfect. Sometimes we may get it wrong. If you are 18 or older and you were put into the under-18 experience by mistake, you can verify your age.
Earlier versions of ChatGPT relied on fixed content policies that applied the same standards regardless of user age, leaving end users or third-party applications responsible for managing age-gating. The updated approach places initial filtering directly at the model level, potentially reducing the risk of minors accessing inappropriate content through generative AI.
According to the parent company:
If you are logged in, you can start age verification here: https://chatgpt.com/verify_age
To verify your age:
- Sign in to ChatGPT on the web.
- Open Settings.
- Under Account, check whether you are in under-18 mode, then choose Verify age.
- Follow the steps with Persona.
Persona is a trusted third-party company we use to help verify age.
Depending on your country, Persona may ask for one or both of these:
- Live selfie: Use your phone or webcam to take a real-time photo of your face.
- Government ID: Upload a photo of a government-issued ID (for example, a driver’s license, passport, or state ID). Persona checks your date of birth and that your selfie matches your ID photo. Accepted IDs vary by country.
According to OpenAI, Persona is responsible for handling the identity verification process and deletes any submitted ID or selfie within seven days of completing verification. OpenAI does not receive or store users’ ID documents or selfies; instead, it only receives a date of birth or an age prediction, which is stored securely and used solely in accordance with OpenAI’s Privacy Policy.
OpenAI said its age estimation feature is currently optional and slated for phased rollout, with plans to provide developers and platform partners with greater control over how age-related filters are applied within third-party integrations of ChatGPT. The company also plans to publish transparency reports on how often the system triggers age-based restrictions.
ChatGPT kicked off plans to roll out an automated age-prediction system back in September of last year. The goal was to help the AI chatbot tell the difference between teenagers and adults.
