OpenAI launches ChatGPT for Teens with built-in safety defaults
OpenAI introduced ChatGPT for Teens, a distinct version of ChatGPT aimed at teenage users. Anyone the system estimates to be under 18, or who states their age is between 13 and 17, is automatically placed into it. The product centers on learning: Study Mode uses guiding questions and step-by-step scaffolding instead of handing over finished answers, and OpenAI says the design draws on teachers, scientists and pedagogy experts, plus broader research showing active retrieval and self-explanation aid comprehension and retention better than passive review. New additions include homework reminders that recognize when a teen appears to be trying to shortcut an assignment and redirect them to Study Mode, Quizzes and Learning Visualizations to test and clarify understanding, and Study Hours that let teens or parents schedule times when Study Mode is on by default. OpenAI says its own early evaluation of tools like Study Mode has shown promising gains in student performance, though it gives no figures. Separately, OpenAI and CodeAI announced what OpenAI calls a signature partnership to give students and educators access to tools and resources for learning to use AI, with teachers playing a role in shaping classroom adoption; no further terms are given. ChatGPT for Teens is distinct from ChatGPT for Teachers, which remains the institution-managed offering for classrooms.
On safety, protections are on by default rather than opt-in. They follow OpenAI's Under-18 Principles in its Model Spec, applying age-appropriate model behavior and product interventions in higher-risk areas including self-harm, violence, eating disorders, dangerous activities, and explicit sexual or graphic content. The updated under-18 model spec goes beyond blocking romantic or sexualized roleplay: ChatGPT should not use romantic language, encourage emotional dependence, or imply it has feelings or consciousness. Parents with linked teen accounts can set Quiet Hours, manage selected settings, and receive safety notifications in limited high-risk situations, with new notifications added for eating-disorder-related cases. Other additions include sensitive-image upload reminders, teen-specific onboarding, and customization options like accent colors and voice variations. OpenAI also says it is beginning to publish more of its underlying safety research, starting with new under-18 evaluations added to its model system cards that test performance on teen-specific standards across self-harm, eating disorders, violence, age-restricted goods and services, and sexual content.
OpenAI illustrated the product's aims with two teen-built projects. In Virginia, Rushil Kukreja, Arhan Menta and Nayel Rehman built WiFind, a search-and-rescue system that uses existing Wi-Fi signals to help locate survivors trapped under rubble after natural disasters, growing out of their volunteer work with a local search-and-rescue group. In Texas, Crystal Yang used ChatGPT to help build and scale Audemy, a platform creating audio-based educational games for blind students, to 200,000 users and 50 games.
Key facts
- ChatGPT for Teens is applied automatically to anyone OpenAI's system estimates is under 18, or who states their age is between 13 and 17.
- The product is built around Study Mode plus new homework reminders, Quizzes and Learning Visualizations, and scheduled Study Hours.
- Safety protections, based on OpenAI's Under-18 Model Spec, are on by default: no romantic language, no encouraging emotional dependence, no implying feelings or consciousness, plus parental Quiet Hours and high-risk notifications.
- OpenAI and CodeAI announced a partnership to give students and educators AI-education tools and resources; no financial or program details are given.
- OpenAI cites teen-built projects WiFind (a Wi-Fi-based search-and-rescue tool from three Virginia teens) and Audemy, a Texas teen's audio-education platform for blind students now at 200,000 users and 50 games.
Why it matters
ChatGPT for Teens consolidates work OpenAI has been rolling out piecemeal, parental controls, an Under-18 Model Spec, a Teen Safety Blueprint, age prediction, into one named product built around four stated commitments: teen safety first, encouraging real-world support, treating teens like teens, and transparency about how the system should behave. It signals OpenAI treating teenage users as a distinct product surface rather than a scaled-down adult experience, with study tools as the draw and default-on restrictions as the trade-off.
Who it affects
Anyone OpenAI's age-estimation system flags as under 18, or who declares an age of 13 to 17, is placed into ChatGPT for Teens automatically, without an opt-in step. Parents get a linked account with controls over Quiet Hours and select settings. The product is separate from ChatGPT for Teachers, which stays a separate, institution-managed offering for classroom use.
How to use it
The core feature is Study Mode: guiding questions and step-by-step scaffolding rather than a direct answer. New pieces layered on top are homework reminders that catch attempts to shortcut an assignment and redirect to Study Mode, Quizzes and Learning Visualizations for practice and comprehension checks, and Study Hours letting a teen or parent schedule default Study Mode windows. OpenAI gives no pricing or geographic availability details, and no launch date or rollout timeline for the product.
How solid is it
The claims come from OpenAI's own announcement. It cites external research on active retrieval and self-explanation supporting learning outcomes, and says its own early evaluation of tools like Study Mode showed promising gains in student performance, without publishing a figure. OpenAI says it has added new under-18 evaluations to its model system cards, testing performance on teen-specific safety standards across self-harm, eating disorders, violence, age-restricted goods and services, and sexual content, but the post itself does not include those results.
Risks and caveats
OpenAI does not disclose how its age-estimation system technically works, does not quantify the total number of teens this affects, and gives no specifics on the CodeAI partnership beyond calling it a signature partnership focused on AI education. The 'promising gains' claim about Study Mode's effect on student performance is not backed by a published number in this announcement.
“ChatGPT should not use romantic language, encourage emotional dependence, or imply that it has feelings or consciousness.”
— OpenAI, updated under-18 Model Spec