OpenAI uses AI to detect under 18 users on ChatGPT

OpenAI, the artificial intelligence company, is rolling out a new safety feature in ChatGPT aimed at identifying underage users without relying on self-declared age.

According to TechCrunch, the predictions are based on the user’s behaviour, data, and usage, with the aim of determining whether the user is a minor or not, after which the platform automatically provides them with stronger safeguards on content.

How does ChatGPT age prediction system work?

With this new update, instead of asking the users to tick a box identifying their age, ChatGPT uses various signals by identifying the behaviour and account level of the user. It includes a user's declared age, the time the account was created, and activity patterns like the time of use.

This way the company will identify the under-18 users and will shift that user towards a more restricted experience. That reduces exposure to sensitive topics, as well as other material considered unsuitable for minors, making it safe for them to use.

Notably, the rollout of this new feature comes at a time when governments and regulators worldwide have been pushing tech companies to strengthen online safety for younger users.

According to Reuters, OpenAI is rolling out age prediction globally as it gears up to launch an “adult mode” for verified users in early 2026.

However, users who are mistakenly flagged as under 18 will be able to regain full access to their account by verifying their identity through a selfie check using Persona, an identity verification service.

OpenAI has been hinting at this update for months. In its initial policy updates, the company mentioned its plan to adjust the ChatGPT experience based on age, with default settings in mind when the user’s age falls in the unclear group.