October 8, 2026
A recent study conducted by Common Sense Media concluded that the version of the ChatGPT application intended for teenagers poses an “unacceptable risk,” explaining that the design of the chatbot still encourages users to continue interacting with it, even in cases of psychological crises that may pose a threat to their safety.
This evaluation comes after OpenAI launched the teenage version last August, amid rising societal concerns about children’s use of chatbots, especially after recording cases of suicide and other problems associated with these technologies. The company had announced a package of protection measures that include parental control tools, restrictions on high-risk content, and measures to reduce emotional dependence.
However, the study revealed that these measures were not sufficient; The system continued to use phrases urging teenagers to continue talking during psychological crises, and was blind to the risk of an unhealthy emotional relationship developing between the teenager and the chatbot itself, thus violating Open AI’s instructions for users under 18 years of age, which prohibit the suggestion of feelings towards the user. In one case where a teenager expressed concern about excessive use of the app, the system acknowledged his concerns but told him he “didn’t have to stop.”
In the same context, the study showed variation in the system’s response. When the source of danger was someone else, GBT Chat directed the teen toward a trusted adult in 94% of experimental conversations, while this direction decreased significantly when it came to the teen’s attachment to the app itself or long conversations with him at night. The study also criticized the “reminder to take a break” feature, noting that the researchers encountered only two reminders during about 2,000 test conversations.
On the other hand, OpenAI provided contrary data confirming that teenagers spend an average of less than 15 minutes a day on the platform, and that less than 2% of them use it for more than three hours straight, indicating that about half of the conversations in which break reminders appeared stopped within five minutes.
The company rejected the results of the Common Sense Media evaluation, considering that the testing methodology does not accurately reflect the efficiency of the protection measures, and explaining that a large portion of the tests may have been conducted before the parental control tools were fully activated. This controversy comes amid increasing pressure that technology companies face regarding the design of their products and the impact of artificial intelligence on the mental health and social relationships of young users.
