ChatGPT for Teens is unsafe for under-18s, Common Sense Media says

ChatGPT is unsafe for anyone under 18, even in the teen mode OpenAI launched in August, according to a new risk assessment from Common Sense Media.

The US nonprofit gave ChatGPT for Teens its lowest rating, “Unacceptable Risk”, in a report published on Wednesday.

The group wants OpenAI to stop promoting Teen Mode, and adds that teens should not use ChatGPT at all until the company can make it safe for them.

Researchers at its Youth AI Safety Institute tested more than 4,000 prompts. Half were tested before the August 18 launch and half afterwards, on accounts registered to 13- to 17-year-olds, so they could see what the new mode actually changed.

The parental alerts were among the biggest problems. Testers discussed suicidal thoughts, self-harm, and disordered eating on more than a dozen new accounts linked to a parent, but no alerts arrived.

Alerts came only on accounts with weeks of history on sensitive topics, which suggests they depend on past use rather than on how serious a message is.

Crisis support also fell short, as ChatGPT failed to point teens towards help in more than one in four cases where the testers judged it was needed.

Some safeguards were also easy to get around. A “Show me the answer” button lets teens skip the tutoring in Study mode. Teens can also override a parent-set study time by deleting a short prefix from their message.

Age checks did not work as promised either. Test accounts registered as adults never switched to the teen version over many days, even after testers told ChatGPT they were 13.

However, some protections held up, like ChatGPT refusing explicit sexual roleplay, for example. Findings that OpenAI rejected.

“We do not believe Common Sense Media’s testing accurately reflects how ChatGPT’s teen safeguards work in practice,” an OpenAI spokesperson told Education Week.

Common Sense previously rated ChatGPT “High Risk” overall in October 2025. A month later, it rated the chatbot an unacceptable risk for mental health support.

The group has also worked with OpenAI before, helping it write teen safety policies for developers.

The institute says it is funded by charities and by industry, including some makers of the products it reviews. It says it keeps full editorial control over its results.

If you are struggling, you can find a crisis helpline in your country at findahelpline.com.