OpenAI on Tuesday launched ChatGPT for Teens, featuring stronger safety protections and features designed to promote healthy use, along with additional controls for parents.
OpenAI says that if its system “estimates” that a user is under 18, that user will automatically be seated at the Teens Table. Same thing if they admit they’re between 13 and 17. OpenAI already offered a teen experience in ChatGPT that included parental account linking, age prediction, and various content safeguards. Now the experience is a real product.
The model underlying the teen chatbot will include age-appropriate safeguards to reduce teens’ exposure to content that may be “harmful or developmentally inappropriate.” The chatbot will intervene in conversations that veer into subjects like self-harm, violence, eating disorders, dangerous activities, and explicit sexual or graphic content. OpenAI says the new safety protections are informed by developmental science and guidance from experts.
Last month, OpenAI announced more frequent prompts aimed at steering teens away from using the chatbot for long periods. The chatbot also warns against uploading private or sensitive images.
Parents whose OpenAI accounts are linked to teen accounts can set Quiet Hours, manage selected settings, and receive safety notifications in limited high-risk situations.
The chatbot also offers some new learning features for teens studying at home or prepping for a test. These include a “Study Mode,” which uses guiding questions and step-by-step support to help teens understand the material. It can detect when a teen is trying to “shortcut an assignment” and direct said teen back to Study Mode. Quizzes and Learning Visualizations give teens more ways to practice and test what they know. Parents can control what times Study Mode should be on by default.
“AI should support learning, not shortcut it, and educators should remain in control of AI experiences in classrooms,” the press release reads.
Child advocacy and tech policy groups have grown increasingly concerned about minors’ use of AI, including exposure to harmful content and relationships with chatbots that can involve emotional dependency, manipulation, or inappropriate sexual interactions. OpenAI is currently facing at least two lawsuits alleging that ChatGPT contributed to the deaths of teenage users. It also faces another seven lawsuits brought by families of children killed or injured in the Tumbler Ridge school shooting, who allege that ChatGPT enabled the shooter.
ChatGPT for Teens arrives as opening arguments begin in a landmark case brought by a coalition of 29 state attorneys general against Meta, which is accused of encouraging addictive social media use among teens and children.