OpenAI uses AI to recognize "children." Is this system reliable?

OpenAI officially deployed an AI age prediction system to identify ChatGPT users under 18. According to the latest news, the system has been launched in the ChatGPT Consumer Plan and is expected to be rolled out in the EU region in the coming weeks. The emergence of this system reflects the increasing global emphasis on protecting minors in AI applications and also exposes inherent contradictions in technological identification.

How the system works

OpenAI’s age prediction system does not simply rely on the birth date provided during registration but evaluates the user’s true age by analyzing multiple behavioral signals. Specifically, it includes:

• Account duration
• Usage activity patterns
• Daily behavioral characteristics

Once the system identifies a user as a minor, the account will automatically be subject to content restrictions, especially strict controls on access to violent and sexually sensitive content.

New parental tools

In addition to age recognition, OpenAI has also introduced supporting parental control features, including:

• Parents can set quiet hours for their children’s accounts
• The system detects severe emotional distress in children and sends notifications to parents

This means restrictions are not only on content but also on usage time and mental health monitoring.

Risks of misjudgment and remedy mechanisms

The biggest issue with this system is that AI-based age identification inherently carries a risk of misjudgment. Adult users may be mistakenly identified as minors due to unusual behavior patterns (such as irregular usage times or queries leaning toward educational content).

OpenAI’s solution is to introduce a third-party verification service, Persona. Misclassified adult users can upload a selfie or identification documents to verify their identity and restore normal access. Although this process provides a remedy channel, it also involves users uploading biometric information to a third party, raising privacy concerns.

Why launch now

On the surface, this appears to be OpenAI’s emphasis on protecting minors. But in deeper terms, there are several background factors worth noting:

1. Regulatory pressure. Regions like the EU are increasingly strict about protecting minors in AI applications. Not implementing corresponding measures could pose compliance risks.

2. Business considerations. According to relevant reports, OpenAI is currently under financial pressure—despite annual revenues exceeding $20 billion, high computing costs cast doubt on profitability. In this context, investing in compliance costs is a necessary, albeit forced, choice.

3. Trust restoration. Recently, Elon Musk sued OpenAI for $134 billion on charges of “fraud,” with core disputes over OpenAI deviating from its original non-profit mission. Launching youth protection measures can be seen as a way to demonstrate that OpenAI still cares about social responsibility.

Key questions

While the system aims to protect, it also raises several questions:

• Can AI accurately identify age? What misjudgment rate is acceptable?
• Is requiring users to upload biometric information truly safer than age restrictions alone?
• If the system becomes overly sensitive, will it affect the normal experience of adult users?

Summary

OpenAI’s AI age prediction system is a pragmatic compromise: it aims to meet regulatory requirements while minimizing impact on adult users. The technology itself is not novel; the key lies in execution effectiveness. The rollout in the EU over the next few weeks and user feedback data will more clearly reveal the system’s true reliability. For OpenAI, this is not just a technical deployment but also a proactive gesture amid a trust crisis.