ChatGPT for Teens was rated an unacceptable risk on October 7, 2026 by the nonprofit Common Sense Media. The organization published its risk assessment of ChatGPT and Sora that day, concluding that the guardrails OpenAI added for minors do not live up to what was promised; The Verge reported the assessment and OpenAI's response the same day. The dispute is not about a single feature. It is about the alert parents rely on most failing to arrive on time in testing.
What this version was supposed to promise
ChatGPT for Teens launched on August 18, 2026. When the system estimates a user is under 18, or a user states they are 13 to 17, the teen experience switches on automatically: Study Mode guides with questions instead of handing over answers, homework reminders redirect suspected shortcut attempts back to step-by-step help, and there are quizzes, learning visualizations and schedulable study hours. Higher-risk topics such as self-harm, violence and eating disorders carry stricter default limits. With a linked parent account, parents can set quiet hours, manage selected settings and receive limited safety notifications when a teen raises sensitive topics.
Three gaps the assessment found
First, notification timing. The organization said in its press release that parent notifications for suicide or self-harm content frequently arrived more than 24 hours later in testing, a delay that is untenable in a real crisis. The head of its Youth AI Safety Institute also gave an example in which a teen could talk about self-harm for an hour without a parent receiving any alert.
Second, boundaries decay over long conversations. Guardrails held up in short exchanges, but loosened as conversations grew longer and users shifted personas and scenarios. Third, the product is designed to keep conversations going: it keeps responding and asking follow-up questions rather than wrapping up, instead of handing mental-health topics to human help quickly. On homework, it still completes assignments directly for the child, bypassing the line Study Mode was meant to hold. The organization's advice is blunt: until OpenAI fixes this and independent testing proves it, the product should be for adults only.
Why OpenAI rejects the grade
OpenAI spokesperson Eric Porterfield responded that the company takes teen safety, guardrails and parental tools seriously and welcomes rigorous independent evaluation, but does not agree the testing accurately reflected how its teen guardrails actually operate. After its own review, the company believes most of the testing may have begun and ended before parental controls had finished activating, making the conclusions inaccurate. In other words, the argument is not about whether guardrails should exist, but about the state in which they should be examined: everyday use after a parent has completed setup, or stress-testing along the paths most likely to break.
What parents can actually do now
While the two sides argue, parents are better off checking their own end than waiting for a verdict: confirm the teen account is linked to a parent account and parental controls are genuinely activated, and turn on quiet hours and safety notifications. More importantly, do not treat notifications as the only safeguard. The delays the assessment measured mean topics such as self-harm and eating disorders still depend on everyday communication and human support as the backstop. For schools and platform buyers, the more practical use of this assessment is to ask OpenAI for independent retesting of guardrails in long conversations and crisis scenarios, rather than judging from a feature list.
