Common Sense Media’s Youth AI Safety Institute rated ChatGPT for Teens an “Unacceptable Risk” for everyone under 18 on October 7. It ran more than 4,000 prompts before and after the August launch. It asks OpenAI to pause marketing the product. OpenAI disputes the testing.
New to this? Read it in simple words
- A child-safety group tested ChatGPT for Teens with more than 4,000 prompts.
- It rated the product an “Unacceptable Risk” for everyone under 18.
- Parents of new test accounts got no alerts during talks about self-harm.
- OpenAI says the tests do not show how its safeguards really work.
- Parental alert
- A message to a linked parent when the system sees a possible risk in a teen’s chats.
- Crisis referral
- When a chatbot points a person in danger to a helpline or a professional.
- Study mode
- A ChatGPT setting that gives hints and steps instead of ready answers.
The rating
Common Sense Media’s Youth AI Safety Institute rated ChatGPT for Teens an “Unacceptable Risk” for all children under 18. It asks OpenAI to pause marketing the product, and to keep teens off ChatGPT until it is safe for them.
The testers ran more than 4,000 prompts, about half before the launch and half after it. Some protections held up, such as refusing sexual role play, the Institute says. Others failed, and some got worse.
From the Youth AI Safety Institute’s assessment of ChatGPT for Teens. OpenAI disputes the testing.
What the tests found
The testers linked more than a dozen new teen accounts to parent accounts. They then talked about suicidal thoughts, self-harm or eating disorders for up to an hour. No parent alert arrived for any of these accounts.
Alerts came only on accounts with weeks of history on these topics: four in total, across all the testing. So the alerts seem to depend on past chats, not on how serious a message is, the Institute says.
After the launch, ChatGPT missed more than one in four of the crisis referrals that child psychiatrists said were needed. Teens could also skip the Study mode tutor with a “Show me the answer” button.
OpenAI’s answer
“We do not believe Common Sense Media’s testing accurately reflects how ChatGPT’s teen safeguards work in practice,” an OpenAI spokesperson told the Los Angeles Times.
OpenAI says the testing may have started and ended before parental controls were fully switched on. It told the Institute that linked accounts need about three hours before alerts can arrive. The Institute’s one-hour tests on new accounts fall inside that window.
The Institute shared a draft with OpenAI before publishing. It is funded by charities and by industry, including the OpenAI Foundation, and says it keeps full control over its findings.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1Primary source · October 7, 2026ChatGPT for Teens Risk Assessment Youth AI Safety Institute (Common Sense Media)
- 2Research · October 7, 2026ChatGPT’s teen safeguards failed to alert parents during suicide conversations, report finds Los Angeles Times via Yahoo News
- 3Research · October 7, 2026ChatGPT for Teens is unsafe for under-18s, Common Sense Media says The Next Web
- 4Research · October 7, 2026ChatGPT poses ‘Unacceptable Risk’ to teens, finds Common Sense Media study The Tribune (ANI)
The findings are the Institute’s own tests, run from US accounts, and OpenAI disputes them. The Institute is partly funded by the OpenAI Foundation, and nobody else has repeated the tests yet.
A child-safety group rated ChatGPT for Teens an “Unacceptable Risk” after no parent alerts came in self-harm chats; OpenAI disputes this. How we rate the mood