ChatGPT's age prediction reopens the debate on privacy and child protection

OpenAI has started to deploy in ChatGPT a function that attempts to estimate whether the user is older or younger with the stated objective of preventing young people from accessing adult or dangerous content. Instead of relying only on the date of birth that the user has introduced when registering, the system uses an age prediction model that observes signs such as the topics you start in conversations and use patterns, for example the time slots in which you use the service.

The idea behind this additional safety layer is simple: to reduce the risk of minors reading harmful instructions, dangerous viral challenges or content that promote extreme diets and harmful physical stereotypes. OpenAI explains that these predictions are used to decide whether to apply an experience with restrictions designed for users under 18 years of age, and that these filters limit access to high-risk content.

OpenAI recognizes, however, that the system is not infallible. The public notice of this function itself states that the model may be wrong and may classify an adult as a minor if its behaviour resembles that of a teenager. If this happens and you are of age, you can go through a verification process to restore the experience without restrictions. The company has published details on the operation and limitations of age prediction at its own help center: official explanation of OpenAI.

To verify the age of majority OpenAI uses an external supplier, Person, who does the verification by comparing a selfie in real time with an official document (identity card, passport, driving licence, according to country). OpenAI ensures that these data (selfie and image of the document) are deleted by Person within seven days of verification. If you want to start the check when you are connected, the company indicates that the access to the check is available at: https: / / chatgpt.com / verify _ age and in your supplier you can consult more information at the Personal website.

Beyond the intention - to protect minors - such measures raise relevant questions about privacy, precision and bias. The use of indirect signals (conversation issues, hours) to infer age opens the door to systematic errors: people from different cultures, shift workers or users with unconventional interests could be misclassified. In addition, any system that uses biometric features to verify age is placed on a sensitive ground; organizations such as the Electronic Frontier Foundation have warned about the risks associated with the use of biometric data and their potential abuse, as well as the need for clear limits on their collection and storage ( more about biometry and privacy).

From a regulatory point of view, the measure also interacts with existing laws on child protection and personal data. In the United States, for example, the Law on the Protection of Online Child Privacy (COPPA) requires specific safeguards for data for children under the age of 13, and in the European Union the General Data Protection Regulation (GDPR) requires legal bases and transparency when dealing with sensitive data or automated decisions affecting people. If you are interested in the basic legal framework around child protection and network privacy, you can consult resources such as the COPPA explanation on the Federal Trade Commission website: COPPA (FTC) and a practical guide to the GDPR: GDPR.

Another point to consider is confidence in third parties that manage verification. Although OpenAI indicates that Person eliminates images in a short time, these promises require audit and transparency to be truly credible. Clear communication on what is shared, how long it is held and who can access such data is essential for users to be able to assess risks before verifying their identity with a photo and a document.

For users there are immediate practical implications. If you are a minor, you will notice limitations on certain issues that were previously available; if you are an adult and the system classifies you as a minor by mistake, you will have to go through age verification to recover full access. In both cases, any automated decision should be accompanied by clear information on why a restriction was applied and how to request a review.

On the social and ethical level, the deployment of models that infer sensitive characteristics raises questions of proportionality: to what extent is it reasonable for a company to investigate patterns of use to deduct age? What margin of error is acceptable when the objective is to protect minors without creating unduly invasive barriers for adults? Public discussion and independent monitoring are necessary to balance the protection of young people with privacy rights and the risk of discrimination.

In short, ChatGPT's new age prediction function aims to reduce damage by limiting access to hazardous content by minors, and offers verification ways to correct errors. But their effectiveness and legitimacy will depend on the quality of predictions, on how biometric data are protected and on transparency against users and regulators. If you are concerned about how this will affect your use of ChatGPT, check the official explanation of OpenAI, assess the risks before checking with documents and keep an eye on independent regulatory developments and analysis that assess the actual impact of these measures.