ChatGPT adds a teen mode — parents can limit it, but not read it

ChatGPT now behaves differently for teenagers. OpenAI launched ChatGPT for Teens on August 18, and the system is automatic: users who are thirteen to seventeen years old get a version of the product that is already configured with tighter limits than the adult default. No action is needed from the teenager, and no opt-in is required from a parent. The feature started rolling out globally that week and is expected to complete its two-week rollout to all markets.

The parental controls, accessible through a family account, are more thoughtful than most platforms offer. A parent can set quiet hours — specific windows of the day when ChatGPT is inaccessible to the teen account — and independently disable voice mode, image generation, or the persistent memory feature that lets ChatGPT remember details across conversations. Each of these is a separate toggle. There is also an option to opt out of OpenAI using the teen’s conversations to train future models, something not enabled by default on standard accounts.

What parents cannot do is read what their child has typed to ChatGPT. This is a deliberate design decision, not a gap in implementation. OpenAI’s position is that teenagers who suspect their conversations might be shared with parents will simply not use the product for the things they actually need help with — and a teenager who cannot ask an AI about something sensitive, without judgment, loses the most useful part of the technology. The controls give parents tools for setting limits around time and mode; the content of the conversations remains private.

The model itself has built-in restrictions that apply regardless of parental settings. ChatGPT for Teens has stricter handling of topics around self-harm, eating disorders, and violence. The model is trained to redirect rather than engage on those subjects, and the same restrictions apply in Study Mode — a new feature that turns responses into quizzes and practice questions rather than direct answers. Study Mode is designed to support homework without replacing it, requiring the teen to retrieve information rather than simply copy it.

The question of age verification is where the system’s limits become apparent. OpenAI uses an age-prediction model, not a formal identity check, to classify users as teenagers. Teenagers who provided a false birth date when first signing up — a common workaround on age-restricted platforms — will likely not be reclassified. The rollout is aimed at new accounts and accounts with correctly registered birthdates. OpenAI has not disclosed what proportion of its existing user base falls into the 13-to-17 demographic, but researchers have consistently found that the group is large and disproportionately heavy in its use of AI tools for academic purposes.

The broader design logic here is worth pausing on. OpenAI is trying to hold two things at once: enough safety architecture to satisfy parents and regulators, and enough privacy to be genuinely useful to teenagers. The quiet hours and topic restrictions address the first; the refusal to share chat logs addresses the second. Whether that balance is correct is a question reasonable people can disagree about. What is notable is that OpenAI made the tradeoff explicit — they chose teen utility over parental surveillance, and they said so.

ChatGPT for Teens is available now on all platforms. The two-week global rollout began August 18.

Tags: artificial intelligence, OpenAI, chatgpt, ChatGPT for Teens, parental controls, teen safety