Placement in the teen edition rests on the age a user states, or on OpenAI's estimate of it — the company runs no age check.
OpenAI introduced ChatGPT for Teens this week, a separate edition of the chatbot for 13- to 17-year-olds in which curbs on suicide and self-harm, graphic violence, depictions of self-injury and roleplay of a sexual or romantic kind are in force from the first message. It also carries tools meant to encourage balanced use and extra controls for parents.
Anyone who gives an age in that range, or whom OpenAI's system judges to be under 18, is routed there by default rather than into the standard product. There is no outright age check: the company leans on age-estimation technology to gauge from signals — among them the kinds of things someone asks about — whether a user is likely under 18, while its published age rules bar anyone under 13 from the service. How hard the new systems will be to work around, for teenagers who decide not to play along, stays unclear until they are put through tougher trials; young people are generally adept at slipping past parental controls and other attempts to fence in what they do online.
The revised under-18 model spec also takes up emotional attachment, saying ChatGPT should steer clear of romantic phrasing, should not foster emotional dependence and should not suggest it feels anything or is conscious, the stated aim being to strengthen healthy ties in the real world. OpenAI says those safeguards rest on the Under-18 Principles in its Model Spec, drawing on developmental science and expert advice. Allison Mishkin, the company's head of child development, told the Guardian her team identified hypothetical cues a model could give that might lead a teenager to form a relationship with it. A 2025 study by Common Sense Media found that more than 70% of American teenagers had turned to AI chatbots for company, and that half used AI companions regularly.
Parents whose accounts are linked to a teen's can designate quiet hours, periods when the teenager is shut out of ChatGPT, and can be alerted about safety in a narrow set of high-risk cases, such as a user who might hurt themself. Further alerts tied to eating disorders are coming, with the range of information passed on kept deliberately narrow and focused on the points where offline help could count most. None of it applies unless the teenager and a parent or guardian both agree to switch it on: Ann O'Leary, OpenAI's vice-president of global policy, told the newspaper the goal is to make sure the teen product is safe even for teens whose parents do not use parental controls.
OpenAI's announcement casts the edition as one built to help teenagers study, reason critically and grow comfortable working with AI. For schoolwork it carries Study Mode, which leads teens through material with prompts and stepwise help so they grasp the subject rather than being handed answers. New "responsible homework reminders" pick up signs that a teen is trying to take a shortcut on an assignment and steer them into Study Mode to work through it one stage at a time, while Study Hours lets teenagers or their parents designate periods in which Study Mode is switched on automatically.
The rollout lands amid growing public concern over what AI tools do to younger users, and as other services bring in age checks of their own along with safeguards written specifically for teenagers. Much of what arrives here, though, has been around before: age-prediction shipped early in the year, parental controls and study mode turned up about a year ahead of this launch, and a month before it the company said teenagers would get break reminders more often. That leaves the release mostly a gathering of existing protections in one place, with a few modest add-ons and adjustments.