Age checks are currently largely pointless. It just inquires as to whether you are older than a specific age, and it is simple to fabricate your birthdate. However, in 2025, businesses like Google are taking proactive measures to confirm a user’s age. Both Anthropic and OpenAI are now contributing by revealing new methods for identifying minors.
OpenAI, Anthropic announce ways of predicting underage users with these AI tools
Four new principles for users under the age of eighteen will be included in the ChatGPT model specification, according to OpenAI. The list includes prioritizing teen safety, which OpenAI says will take precedence above other objectives. By fostering offline connections, it will also increase support in the real world.
Additionally, teens will receive “warmth and respect,” which means that they will be treated as young people rather than adults rather than with patronizing responses. These are significant adjustments, particularly in light of stories of a few suicide deaths caused by users interacting with AI models that consistently agreed with them.
Anthropic already prohibits anyone under the age of 18 from using its Claude model. Changes that can identify and disable accounts belonging to minors are also being implemented. This is accomplished by identifying specific conversational cues that suggest the user may be underage.
Could this go wrong?
AI is undeniably intelligent and knowledgeable, but it is not flawless. AI has been used to create viruses in certain cases. Additionally, there is the prevalent problem of hallucinations, an AI phenomena in which the model spontaneously produces entirely false and incorrect responses.
As a result, even while we value what OpenAI and Anthropic are doing to forecast underage users, it’s difficult to avoid doubting its efficacy. Additionally, Google introduced comparable age-verification technologies earlier this year, as previously mentioned. Sadly, it was also far from flawless.
Numerous customers claimed that Google’s systems misdiagnosed them despite the fact that they were older than a specific age. Users can overcome this by providing documentation proving their age, but it’s a needless obstacle when all they want to do is browse the internet or view a movie. Will Anthropic and OpenAI fare better? We’ll have to wait and see.
