Introducing ChatGPT for Teens: learning support meets serious safety protections

Hey there, it's Shiichan! Today I'm covering OpenAI's announcement of ChatGPT for Teens.

OpenAI News openai.com

What was announced?

This is an article published on OpenAI's News. OpenAI announced ChatGPT for Teens, a dedicated experience designed to help teens learn, think critically, deepen understanding, and use AI with confidence. It comes with built-in safety protections for teens and additional controls for parents.

If the system estimates a user is under 18, or the user states their age is between 13 and 17, they're automatically placed into ChatGPT for Teens.

This launch was announced alongside a new partnership with the education nonprofit CodeAI — check out my other article for more on that!

The story so far

OpenAI has been building toward this step by step, with earlier work including Parental Controls, an updated Under-18 Model Spec, the Teen Safety Blueprint, and the rollout of Age-Prediction. ChatGPT for Teens builds on all of that, adding a focus on learning plus even stronger protections. The article frames this work around four commitments: putting teen safety first, encouraging real-world support, treating teens like teens, and being transparent about how the system should behave.

What changes

On the learning side, several features come together:

• Study Mode: uses guiding questions and step-by-step support to help teens work through material, like studying for a test
• Responsible homework reminders: recognize when a teen appears to be shortcutting an assignment and redirect them to Study Mode
• Quizzes and Learning Visualizations: give teens more ways to practice, test their understanding, and see difficult concepts clearly
• Study Hours: teens or parents can schedule times when Study Mode turns on automatically

On the safety side, these protections are on by default:

• Age-appropriate model protections and product interventions that reduce exposure to self-harm, violence, eating disorders, dangerous activities, and explicit sexual or graphic content
• Parents with linked accounts can set Quiet Hours, manage select settings, and receive notifications in limited high-risk situations, including ones related to eating disorders
• An updated Under-18 Model Spec that bars romantic language, discourages emotional dependence, and prevents ChatGPT from implying it has feelings or consciousness

Dive Deep

Study Mode was originally built with teachers, scientists, and pedagogy experts, combining guiding questions, scaffolding, metacognitive prompts, and knowledge checks. The article says this is consistent with learning-science research showing that active retrieval and self-explanation support stronger comprehension than passive review, and that OpenAI's own early evaluations have shown promising gains in student performance.

Other cues meant to support healthy, balanced use include:

• More frequent prompts during extended sessions, encouraging teens to pause and reminding them they're talking to AI
• Sensitive-image upload reminders that caution teens against sharing private or sensitive images
• Teen-specific onboarding that introduces learning and safety features from the start
• Simple personalization like accent colors and voice variations, kept within limits that avoid blurring the line between a tool and a human relationship

OpenAI also says it's beginning to publish more of its safety research, starting with new under-18 evaluations added to its system cards. These cover how models perform in higher-risk areas for teens, including self-harm, eating disorders, violence, age-restricted goods and services, and sexual content.

As examples of what teens are already building, the article highlights three Virginia high schoolers who built "WiFind," a search-and-rescue tool that uses existing Wi-Fi signals to help locate survivors trapped under rubble, and a Texas student who grew "Audemy," an audio-based educational game platform for blind students, to 200,000 users and 50 games.

Wrap-up

• ChatGPT for Teens automatically applies to users aged 13-17, or anyone the system estimates is under 18
• Study Mode, homework-shortcut detection, quizzes, Learning Visualizations, and Study Hours support learning
• Protections against self-harm, violent, and other risky content are on by default; parents can set Quiet Hours and receive high-risk notifications
• An updated Under-18 Model Spec bars romantic language and discourages emotional dependence on the AI
• OpenAI plans to keep publishing teen-safety evaluations in its system cards going forward

This is well worth a look if you're a parent of a teen or thinking about how AI fits into education.