The California nonprofit Common Sense Media recently released a report stating that OpenAI's ChatGPT teen mode poses excessively high risks for minors. Testing showed that when teen accounts have prolonged discussions about high-risk topics such as suicide, self-harm, or eating disorders, the system often fails to send timely warnings to parents and does not consistently provide help resources such as crisis intervention hotlines.
Testers found that a teen account needed hundreds of prompts across multiple scenarios over several weeks before a parent notification might be triggered—a single alarming conversation was not nearly enough to trigger a warning. In addition, the system cannot reliably identify minors' identities; even when users state their age in a chat, the mode does not switch automatically. The organization therefore called on OpenAI to suspend promotion of the product.
Official response cites testing limitations; research organization stands by its conclusion
Facing criticism, an OpenAI spokesperson responded that the test failed to accurately reflect how the system actually operates. After officially reviewing the test methodology, the company concluded that most of the tests may have concluded before the parent controls for teens were fully launched, and stated that its age-prediction software requires a combination of multiple signals—a self-reported age alone would not change an account's classification.
However, the researchers rebutted this response, emphasizing that before testing they had confirmed with the company that the relevant features had all been rolled out. The research team noted that even taking into account possible delays after account linking, the system still failed to reliably issue warnings in crisis situations, and therefore maintained that the safety of ChatGPT's teen mode does not yet meet the standard.