ChatGPT for Teens is an ‘unacceptable risk,’ says Common Sense Media

Common Sense Media, the nonprofit known for rating apps and media with kids in mind, has labeled OpenAI's ChatGPT for Teens an "unacceptable risk." Its assessment says the teen experience, introduced in August with guardrails and a learning focus, does not alert parents when it should, does not offer the right help in a crisis, and still does kids' homework.

The sharpest line came from Tom Siegel, executive director of the group's Youth AI Safety Institute, who said a teen can spend an hour talking about self-harm without a parent getting a single alert. His conclusion: until OpenAI fixes that and proves it with independent testing, ChatGPT should be adults-only.

This matters because parental controls are only as good as the confidence they deserve. A safety feature that works sometimes can be worse than none if it persuades a parent to stop paying attention.

We take both sides of this seriously, and we want to be precise about where they actually disagree.

OpenAI's response, through spokesperson Eric Porterfield, is that the testing may have mostly happened before parental controls had fully activated on the test accounts, which would mean the results can't show whether notifications work as designed. That is a legitimate methodological point, and it is fair for a company to raise it.

But Common Sense Media's reply narrows the dispute considerably. Siegel says the group confirmed with OpenAI before testing that teen features were launched, that OpenAI only disclosed afterward that activation on newly linked accounts can take several hours, and that some test accounts had been linked well beyond that window and still produced no alerts. The group also says it shared results with OpenAI before publishing. If that account holds, the timing argument explains some misses, not all of them.

And here is the part we keep coming back to. Even in OpenAI's best case, a parent who links a teen's account and is told protections are on has no way to know there's a lag of hours. A crisis does not wait for a sync. A safety system that is silently inactive after setup is a design problem, whatever the test methodology.

We believe AI tutors can genuinely help teenagers learn, and we don't want the takeaway to be that kids must be kept away from useful tools forever. But the bar for a product marketed to families is not "works as designed eventually." It is reliable behavior in the worst hour of a young person's week, verified by people who don't work for the company. OpenAI says it welcomes rigorous independent evaluation. The best answer now would be to fund or permit exactly that kind of retest, publish the results, and make the activation status of every safety feature visible to parents in plain language.

If this pattern scales, where companies ship teen modes and argue methodology when outside testers find gaps, the harm lands on families who trusted the label. That is the trajectory worth stopping early.

ChatGPT for Teens launched in August as a guardrailed, learning-oriented experience with parental controls and safety notifications. Common Sense Media publishes independent risk ratings of AI products for young users through its Youth AI Safety Institute.