OpenAI wants to analyze chats to guess the user’s age and report dangerous cases to parents or even authorities. The debates began: Is this really the first step to parental control or to the beginning of the end of privacy?
Sam Altman, CEO of OpenAI, admits in his post: “Some of our principles are in conflict,” talking about prioritizing teen safety over privacy and freedom. Still, Sam “doesn’t expect that everyone will agree with these tradeoffs.”
The point is OpenAI crafts the system to understand whether a ChatGPT user is over or under 18. People classified by AI as teens (even if this is untrue) will automatically be directed to the under-18 experience with blocked graphic sexual content and reporting to the parents or the authorities in case of harmful hints like suicidal ideation.
If you worry that the system mistakes you for a teen, you may be asked to provide your ID as proof to unlock adult capabilities.
Let us recall the background and reason for such OpenAI activity towards teens’ safety:
In April 2025, Adam Raine, a 16-year-old from California, died by suicide after months of interactions with ChatGPT. His parents have filed a lawsuit claiming the system provided him with details about suicide methods, helped draft notes, and advised secrecy from family members.
The case alleges wrongful death and design flaws against OpenAI. OpenAI has said it is saddened by the incident, emphasizing that its products include safeguards and crisis resources, but acknowledged these measures can sometimes fail, particularly in extended conversations. The company began to develop parental controls, with plans to implement them by the end of September 2025.
While the intentions are good, as Sam Altman expected, this news sparked debates over users’ privacy:
What? HOW!?
— joshua (@4xiom_) September 16, 2025
What is next, contacting employers or authorities on your users for their private chats? Now that you have them ID’d then the gates are open to what is deemed to be “dangerous activity”. This a tremendous breach of privacy usage. I understand the intent but this… pic.twitter.com/fXf296yYgd
Yuvraj user highlighted that the statement about just “developing advanced security features to ensure your data is private, even from OpenAI employees” is already scary.
"In some cases or countries we may also ask for an ID; we know this is a privacy compromise for adults but believe it is a worthy tradeoff."
— Peter Gallagher (@pwgallagher) September 16, 2025
This is bad and foolish. Govt. controls on speech are not about speech (or parenting) but about control. OpenAI will now provide a lever. https://t.co/Xemz0UfY29
The fact that OpenAI is transparent about trade-offs (privacy vs safety) is a good sign: at least they admit there are conflicts instead of denying them. It helps with trust.
The default-to-under-18 approach can help with safety, but it also heightens the risk of misclassifying adults. How the “age-prediction” works, its accuracy, and how often mistakes happen will matter a lot.
Meanwhile, asking for ID when disputed adds friction and has its own risks: people may not want to share documents.
At HeyLocate, we agree that parental controls are useful, but they depend a lot on how they’re designed: are they heavy-handed, do they respect teen autonomy, do they allow teens to reach help without fear of disclosure?
The strategy depends heavily on trusted, effective detection and classification. If the AI misfires in high-stakes cases, it could cause harm instead of preventing it.
Monday-Sunday: 5:00am - 8:00pm (PT)
Leave a Comment