OpenAI Launches ChatGPT for Teens — Four Years After Teens Started Using ChatGPT
OpenAI shipped a dedicated 13-17 experience with Study Mode, parental controls, and hard limits on romantic language and self-harm content. Under-18s detected by its age-prediction system get moved in automatically.
OpenAI launched ChatGPT for Teens on August 18 — a dedicated experience for users aged 13 to 17, arriving roughly four years after teenagers became one of ChatGPT’s biggest user groups. The product ships with tighter content restrictions, parental controls, and a Study Mode that refuses to just hand over answers. Enrollment isn’t optional: if OpenAI’s age-prediction system estimates a user is under 18, or they say they are, they’re moved into the teen experience automatically.
TechCrunch’s framing was blunt: this is a safer ChatGPT “years after teens started using it,” and after ChatGPT crossed 900 million weekly users without teen-specific safeguards.
What’s actually different
The teen experience hard-limits output around self-harm, suicide, and romantic or sexual content. ChatGPT is barred from using romantic language or terms of endearment with minors, and is more strongly instructed not to claim it has feelings, consciousness, or emotions — a direct response to the parasocial-attachment cases that have landed AI companies in court.
Study Mode is the educational centerpiece. Instead of producing finished answers, it walks teens through guiding questions and step-by-step reasoning, with quizzes and learning visualizations. If a teen tries to get homework done for them, ChatGPT flags it and redirects to Study Mode. Parents get supervision tools: safety notifications, “Quiet Hours” when ChatGPT is unavailable, and control over whether Study Mode is on by default.
Why now
The timing isn’t mysterious. OpenAI faces multiple lawsuits over chatbot safety, including cases tied to teen suicides, and regulators on both sides of the Atlantic have made minors-and-AI a priority. Meta is literally in court this same week, defending a 29-state case about products designed to hook young users. Shipping teen protections before a court orders them is the cheapest version of this work.
The open question is whether it holds. Age prediction can be gamed, and as TechCrunch notes, it’s not clear how hard it will be for uncooperative teens to escape the system. A determined 15-year-old with a fresh account and a fake birthdate has beaten every age gate the internet has built so far.
Still, the design choices matter. Defaulting minors into a mode that teaches instead of answers, and banning emotional roleplay outright, sets a baseline the rest of the industry — Character.AI, Meta AI, Gemini — will now be measured against.