In response to growing concerns about artificial intelligence's effects on youth, OpenAI has launched an "age prediction" capability within ChatGPT. This new tool aims to detect underage users and apply appropriate content limitations to their interactions.

OpenAI has faced significant scrutiny in recent years over ChatGPT's potential influence on children. The chatbot has been connected to several teenage suicides, and, similar to other AI providers, the company has been criticized for permitting discussions of sexual subjects with young individuals. In April of last year, OpenAI had to resolve a software flaw that enabled the chatbot to produce erotic content for users under 18.

The company has been addressing the issue of underage usage for a while, and this latest "age prediction" feature builds upon existing safeguards. According to a blog post published on Tuesday, the new system uses an AI algorithm to analyze user accounts for specific "behavioral and account-level signals" to help identify minors. These signals encompass factors such as the age provided by the user, the account's creation date, and its typical activity patterns.

OpenAI already employs content filters intended to block conversations about sex, violence, and other potentially harmful topics for users under 18. When the age prediction system flags an account as belonging to a minor, these filters are automatically activated.

If an account is incorrectly classified as underage, there is a process for users to verify their adult status. They can submit a photograph to Persona, OpenAI's identity verification partner, to restore full account access.