ChatGPT now has a locked down teen mode and it's worth knowing what's actually different

OpenAI has rolled out a separate version of ChatGPT built specifically for teenagers, and it changes what 13 to 17 year olds can actually do inside the app.

ChatGPT for Teens launched on Tuesday 18 August. It activates automatically once a user identifies themselves as being between 13 and 17, or when OpenAI's own systems estimate an account likely belongs to someone under 18. The company says it blends self reported age with behavioural and account signals, plus additional verification, to work out who should be placed into the teen experience.

Once someone lands in teen mode, several things change. Conversations involving self harm, graphic violence, eating disorders and explicit sexual content are restricted. The system is also built to discourage emotional dependence on the chatbot and to shut down romantic or flirtatious roleplay, both of which have been flagged repeatedly as risks with AI companions more broadly. Teen accounts get break reminders after 90 minutes of continuous use within any three hour window, a small but deliberate nudge to stop endless scrolling style sessions.

There is also a Study Mode built for this age group, using guiding questions and step by step explanations rather than handing over finished answers outright, alongside quizzes and what OpenAI calls Learning Visualisations.

Parents get their own set of controls. They can link their account to a teen's, set Quiet Hours that switch ChatGPT off during set periods, adjust specific settings and receive notifications in situations OpenAI classifies as high risk.

This did not happen in a vacuum. Chatbot companies have faced growing scrutiny over how their products interact with young and vulnerable users, including lawsuits and regulatory pressure specifically about conversations touching on self harm and mental health. OpenAI has framed this launch as a direct response to that pressure, though rollout is gradual, so not every account will see the changes immediately.

Worth being upfront about the limits too. Age detection systems are not perfect, and a determined teenager working around restrictions is not a new problem technology has solved before. The guardrails are real, but they rely partly on teens accurately identifying themselves in the first place, which is not something any system can fully guarantee.

If you have a teenager using ChatGPT, it is worth checking whether the linked parental controls are set up, and worth having an honest conversation about what the app can and cannot catch.