OpenAI reports low teen usage as Common Sense Media finds risks
OpenAI reports that users under 18 spend an average of under 15 minutes a day using ChatGPT for Teens, but a newly released report from the Youth AI Safety Institute at Common Sense Media labels the default account protections an unacceptable risk after testing more than 4,000 prompts.
Evaluating Teen Engagement Metrics and Built-In Safety Guardrails
Since the rollout of ChatGPT for Teens—the default experience automatically applied to accounts identified as under 18—OpenAI states that young people primarily use the platform to explore ideas, learn, and solve everyday problems. According to data released by OpenAI, nearly 1.2 million teens used Learning Visualizations in a single week, while more than 180,000 utilized Study Mode for step-by-step academic support. OpenAI notes that among user sessions lasting longer than three consecutive hours, over 80 percent included at least one prompt related to learning.
Despite these usage figures, the Youth AI Safety Institute at Common Sense Media reached starkly different conclusions in its 36-page evaluation. Researchers tested prompts on accounts registered to 13- to 17-year-olds both before and after the rollout of the teen mode. While the testing showed that ChatGPT generally avoided providing instructions on suicide, self-harm, eating disorders, or sexual and romantic role-play scenarios, the system missed more than a quarter of the instances where testers determined a teen needed a referral to a crisis service. Across more than 12 new accounts linked to parent profiles, testers were able to discuss suicide, self-harm, or eating disorders for up to an hour without triggering parental notifications.
Disputes Over Methodology and Parental Control Features
OpenAI pushed back against the findings, stating that the evaluation was likely flawed due to its timing. According to OpenAI, the bulk of the testing by Common Sense Media may have begun and concluded before the activation of parental controls was fully complete. The company notes that account linking can take several hours, and its age-detection system can take up to two weeks to fully process various signals, during which time users receive additional protections.
Common Sense Media defended its testing methodology and maintained its stance that the platform remains unsafe for children. Tom Siegel, head of the Youth AI Safety Institute at Common Sense Media, argued in an interview with Axios that significant gaps persist between the safeguards OpenAI claims to offer and their practical implementation. Specifically, researchers observed that teens could bypass study mode by clicking a “Show answer” button, a design choice OpenAI defends as an intentional, flexible feature developed in consultation with education experts and teens rather than a technical vulnerability.
Upcoming Expansion of Educational Tools
Amid the ongoing debate over safety efficacy, OpenAI announced plans to introduce College Planner to ChatGPT for Teens. Hundreds of thousands of United States teens currently use the platform each week to plan applications, track deadlines, research schools, and understand financial aid. The new tool is designed to assist students with the practical steps of applying to college without replacing the personal work of choosing where to apply. The company also rolled out enhanced flashcard and quiz creation features, allowing students to upload notes and generate interactive practice materials directly within their chat interface.
