ChatGPT Adds New Teen Safety Features, but Parents Still Can’t Read Chats
Nearly nine in ten teenagers who use ChatGPT interact with it every week, mostly for homework, research and staying organized, according to OpenAI.
As more teenagers rely on the chatbot, parents are asking a simple question: How much can they actually see? OpenAI said the new parental features are designed to “give families more ways to support teens while preserving appropriate privacy for their conversations.”
The company expanded its parental safety system on July 13 to notify linked parents in more high-risk situations, including when a teen’s account is banned for violent activity.
Three days later, it introduced break reminders designed to encourage healthier screen habits for younger users. The message is clear: parents get more tools, but not more access to private chats.
What Parents Can Control
Parents or guardians can link their account with a teen aged 13 or older through an email invitation.
Once linked, they can:
- Set quiet hours when ChatGPT can’t be used.
- Turn off Voice Mode.
- Disable Memory so conversations aren’t remembered.
- Remove image generation.
- Prevent chats from being used to train OpenAI’s models.
- Enable Study Mode by default to encourage guided learning instead of direct answers.
OpenAI says parents cannot view chat history or individual conversations, emphasizing that the controls are intended to manage how teens use ChatGPT rather than monitor everything they say. That boundary hasn’t changed.
What Parents Can and Can’t Do
| Parents can… | Parents can’t… |
|---|---|
| Set quiet hours | Read chat conversations |
| Disable Voice Mode | Monitor chats in real time |
| Turn off Memory | View chat history |
| Disable image generation | Search messages |
| Turn on Study Mode | Receive every conversation |
When Parents Are Notified
OpenAI says emergency notifications are intended for situations involving credible signs of immediate harm, while avoiding alerts for ordinary fictional writing, academic discussions or hypothetical conversations.
The system can alert a linked parent if it believes a teen may be experiencing acute emotional distress or if the account is banned for violent activity.
The company says those alerts are designed to avoid triggering over fictional writing, gaming discussions, political debates or hypothetical questions. In rare situations where OpenAI believes there’s an immediate danger and can’t reach a parent, it says it may contact law enforcement.
Why OpenAI Draws the Line
Sonia Livingstone, professor at the London School of Economics and a leading researcher on children’s digital rights, has argued that young people benefit when digital services balance parental involvement with age-appropriate privacy, rather than relying on constant surveillance.
The safeguards rely partly on an age-estimation system that applies stricter protections if OpenAI believes a user is under 18. Those protections include stronger limits around self-harm, graphic violence and other sensitive content.
The approach isn’t perfect. OpenAI hasn’t disclosed how accurate its age-detection system is or how often its distress alerts produce false positives. Teenagers could also avoid the safeguards by using an unlinked account.
For families already talking openly about AI use, however, the new controls provide another layer of oversight without turning ChatGPT into a tool that parents can monitor line by line.
