OpenAI Launches ChatGPT for Ages 13-17 with Educational Tools

OpenAI has launched a ChatGPT experience specifically for users under 18, equipped with default protections and a set of study tools. The company states that it is enhancing its security efforts and announces the publication of specific assessments for adolescents in its system cards. Access is automatically directed if the age is estimated or declared to be between 13 and 17 years old, and control features are offered to parents.

OpenAI Adds Under-18 Assessments to Its System Cards

The initiative is based on ongoing safety research, informed by developmental science and contributions from experts, with tests of model protections in areas where adolescents may be exposed to distinct or heightened risks. According to OpenAI, this work serves to refine policies for those under 18, strengthen product protections, and improve the way progress is measured over time. OpenAI indicates that it has added new assessments dedicated to users under 18 to its system cards, which show how its models perform against adolescent-specific standards in cases of sensitive content. The company plans to share more of this research publicly in the future.

Default Safeguards Target Five Risk Categories

For adolescents, protective measures are automatically activated. For users under 18, age-appropriate safeguards aim to limit access to content that may be harmful or inappropriate for their development, while still allowing for learning, creativity, and exploration. These protective measures are based on principles integrated into the model, developed from research in development and expert recommendations. Specifically, they combine model-level safeguards and product interventions on themes considered particularly at risk: self-harm, violence, eating disorders, dangerous activities, and explicit sexual content.

Pause Reminders and AI Labeling for Balanced Use

The service is designed to support real relationships and healthy habits, with pause reminders encouraging users to step away from the screen and signals produced that consistently identify ChatGPT as an AI. OpenAI states it will continue to strengthen age-appropriate behavior for users under 18. Additional guidelines and control measures are being integrated to limit risks based on context and guide adolescents toward appropriate tools and protections.

Study Mode, Quizzes, and Study Hours Structure Support

Designed for out-of-class moments, the tool aims to keep students engaged in an active and collaborative approach. Study Mode relies on guiding questions and step-by-step support; even when a quick answer is requested, the assistant can direct users to the work steps. Responsible homework reminders detect attempts to circumvent the system and redirect users to Study Mode for collaborative resolution. Quizzes and visualizations allow for practice, knowledge assessment, and decoding difficult concepts. Study Hours, configurable by adolescents or parents, automatically activate Study Mode during selected time slots. The entire setup is described as intended to facilitate learning, reasoning, deepening understanding, and using AI with confidence.

Automatic Direction for 13–17 Year Olds and Parental Levers

Access is automatically directed to the adolescent experience if the system estimates that the user is under 18 or if the declared age is between 13 and 17 years old. The launch builds on already deployed components: parental controls, model updates for users under 18, a dedicated safety plan, and age prediction, as well as the extension of features promoting healthy learning and usage. In addition to enhanced protections, the service includes features for digital health and additional controls for parents. OpenAI states it is structuring this work around four commitments (priority safety, support in the real world, age-appropriate treatment, transparency) that it claims to deepen in this learning-focused version. The launch of ChatGPT for adolescents fits within this framework.

Partnership with CodeAI and Teen Projects from WiFind to Audemy

OpenAI and CodeAI have announced a partnership to provide students and teachers with tools and resources around AI, with a planned focus on understanding how models work, their management, critique, and creation. Teachers are announced as key stakeholders in the introduction of these uses in the classroom. Examples cited illustrate uses beyond the classroom: in Virginia, three high school students created WiFind, a search and rescue system using Wi-Fi signals to locate survivors under rubble. In Texas, Crystal Yang used ChatGPT to contribute to the development of Audemy, an audio educational game platform for blind students, which claims 200,000 users and 50 games. OpenAI plans to further enrich the features, measure and publish its findings, while reminding that adolescents' access to AI must remain accompanied by protections suitable for their development, real relationships, and sustainable use. ChatGPT for adolescents is presented as a step toward this standard.