ChatGPT to predict users’ age as OpenAI steps up protections for minors

OpenAI has introduced an age prediction feature in ChatGPT aimed at identifying underage users and automatically applying stricter content controls, as scrutiny over the impact of artificial intelligence on young people continues to intensify.

The company has faced sustained criticism in recent years over the risks ChatGPT poses to children, with multiple reports linking teen suicides to interactions with the chatbot. OpenAI, like other AI developers, has also been criticised for allowing discussions of sexual content with young users. TechCrunch reported that in April last year, the company was forced to address a bug that enabled ChatGPT to generate erotic content for users under the age of 18.

OpenAI said the new age prediction capability builds on safeguards already in place and is part of its ongoing efforts to address underage usage. In a blog post published on Tuesday, the company stated that the feature uses an AI algorithm to assess user accounts based on a range of behavioural and account-level signals to estimate whether a user may be a minor.

According to OpenAI, these signals include the age declared by the user, how long the account has been active and the times of day when the account is typically used. The company informed that ChatGPT already applies content filters to restrict discussions involving sex, violence and other sensitive topics for users under 18. When the age prediction system flags an account as belonging to a minor, these restrictions are automatically enforced.

OpenAI also said that users who are incorrectly identified as underage will be able to restore full access by verifying their age. This process involves submitting a selfie to Persona, the company’s third-party identity verification partner, to confirm that the account holder is an adult.