OpenAI introduces ChatGPT for teens, an experience designed to help young people between 13 and 17 learn, develop critical thinking, and use artificial intelligence with greater confidence. The system automatically activates this version when it estimates that someone is under 18 or when they state that they are between 13 and 17 years old.
The proposal combines educational tools with stricter safety controls. The idea? To help AI support understanding and practice, rather than turn it into a machine that delivers quick answers without any real learning.
An AI designed for learning, not copying
ChatGPT for teens includes features intended to keep students involved in the process. For example, someone studying for an exam can ask for an explanation, solve an exercise step by step, and check whether they truly understood the topic.
If the system detects that someone is trying to get the direct answer to an assignment, it can redirect them to a guided learning mode. The main features include:
- Study mode: uses guiding questions, gradual explanations, and comprehension checks.
- Responsible homework reminders: identify possible attempts to skip the process and suggest working through the problem step by step.
- Quizzes: make it possible to practice and check how well someone understands a subject.
- Learning visualizations: help represent difficult concepts more clearly.
- Study hours: teens and parents can choose times when Study mode will activate by default.
The logic behind these tools is simple: learning is usually more effective when someone tries to remember, explain, and apply an idea—not just when they read a finished answer. OpenAI says Study mode was developed with teachers, scientists, and education specialists, and that its initial evaluations show promising results in student performance.
AI can support learning, but it should not replace the effort of understanding.
ChatGPT for teens is mainly intended for moments outside the classroom, when a teacher may not always be available. For educational institutions, OpenAI continues to offer ChatGPT for Teachers, with access managed by schools and districts.
A partnership to prepare the first generation of students with AI
OpenAI also announced a partnership with CodeAI to provide tools and resources for students and educators. The goal is not simply to teach young people how to use an app, but to help them understand how artificial intelligence works, ask it questions, direct it, and create with it.
With this approach, teachers continue to play an important role. Technology can open up new possibilities, but educational communities are the ones that must decide how to incorporate it responsibly in the classroom.
From studying with AI to solving real-world problems
OpenAI highlights cases of teens who are already using AI tools to build projects with real-world impact. In Virginia, Rushil Kukreja, Arhan Menta, and Nayel Rehman developed WiFind, a system that uses signals from existing Wi-Fi networks to help locate survivors trapped under rubble after a natural disaster.
The project grew out of their volunteer work with a search-and-rescue organization. AI helped them research, create prototypes, and move faster from an initial question to a tool designed for emergency teams.
In Texas, Crystal Yang used ChatGPT to develop and expand Audemy, a platform that creates educational audio games for blind students. According to OpenAI, the project reached 200,000 users and includes 50 games.
These examples point in an interesting direction: teens are not using AI only to complete homework. They can also use it to learn skills, create products, and address problems in their communities. Of course, they still need proper guidance—and they must verify that the answers are correct.
Protections activated by default
For users under 18, ChatGPT for teens includes measures designed to reduce exposure to harmful or age-inappropriate content. The protections cover higher-risk areas such as:
- Self-harm.
- Violence.
- Eating disorders.
- Dangerous activities.
- Explicit or graphic sexual content.
- Age-restricted products and services.
These measures are based on the principles for users under 18 included in OpenAI’s behavior rules. The company says this work takes adolescent development research and expert recommendations into account.
Parental controls are also receiving new features. When accounts are linked, parents can set quiet hours, manage certain settings, and receive notifications in limited high-risk situations. OpenAI is adding alerts related to eating disorders, aiming to share only the information necessary at moments when support outside the internet could be important.
Less emotional dependence and more support in the real world
One of the most sensitive areas is the emotional relationship between teens and AI systems. ChatGPT for teens will include reminders to take breaks and signals that clearly remind users they are interacting with artificial intelligence.
The updated rules for minors also state that ChatGPT should not use romantic language, encourage emotional dependence, or suggest that it has feelings or consciousness. This goes beyond blocking romantic or sexual role-play: it seeks to prevent the tool from presenting itself as a human relationship.
OpenAI is also adding other features to promote balanced use:
- More frequent notices during extended sessions recommending a break.
- Reminders about the risks of sharing private or sensitive images.
- A teen-specific initial setup with information about learning and safety features.
- Personalization options, such as colors and voice variations, without trying to erase the difference between a tool and a person.
Evaluations specifically for teens
OpenAI says it will continue researching how minors interact with AI and how to measure the effectiveness of its protections. The company has begun publishing new evaluations for users under 18 in its technical safety documents.
These tests analyze model behavior in sensitive cases involving self-harm, eating disorders, violence, restricted products, and sexual content. Making these results public can help researchers, families, and educators better understand both the capabilities and limitations of these systems.
The challenge is significant. A protection can fail, an answer can be incorrect, and no filter can replace the supervision of responsible adults or professional support when a risky situation arises. That is why transparency and the ability to evaluate results will be just as important as the announced features.
ChatGPT for teens puts forward a necessary idea: young people should be able to explore AI, but with tools that respect their stage of development and strengthen their connections to the real world. The question is no longer whether teens will use artificial intelligence, but how to help them learn to do so with judgment, boundaries, and support.