OpenAI brings age prediction to ChatGPT to limit sensitive content for under-18 users: here's how it works

OpenAI has said that it is bringing an age prediction system to ChatGPT to establish if an account belongs to a user under 18 to ‘turn on extra safety settings.’ The San Francisco-based AI startup says its age prediction model uses a combination of ‘behavioral and account-level signals’ such as the age of the ChatGPT account, usage patterns over time and a user's stated age to zero in on their actual age.

After determining that an account is under 18, OpenAI says its new tool will automatically apply additional protections designed to reduce exposure of minors to sensitive content like graphic violence, viral challenges that encourage risky behaviour, sexual or romantic roleplay, depictions of self-harm, content promoting body shaming or extreme beauty standards.

The company says if its system is not confident about someone's age or has incomplete information about a user, it will now default to a safer experience.

‘No system is perfect. Sometimes we may get it wrong,’ OpenAI admits on a support page.

If the user has been incorrectly placed in the under-18 category, they will have a simple way to confirm their age and restore full access using a selfie through Persona via a ‘secure identity-verification service’.

The new system is rolling out globally except for the European Union, where it will ‘roll out in the coming weeks to account for regional requirements’.

Why is OpenAI bringing age prediction to ChatGPT?

OpenAI had announced the age prediction system in September last year when it was under mounting pressure over concerns regarding ChatGPT's impact on mental health and child safety after a series of suicides being linked to the chatbot.

The company is also facing lawsuits from parents of children who have allegedly committed suicide with help from ChatGPT. It is also being investigated by the US Federal Trade Commission (FTC), along with its other peers, on how ChatGPT negatively affects children and teenagers.

OpenAI had also earlier introduced parental controls in ChatGPT which allow parents to link their ChatGPT accounts with that of their children and disable specific high-risk features like voice mode, image generation and memory. Parents also have the option to schedule blackout periods with ChatGPT for their kids and receive proactive notifications when the system detects signs of "acute distress".