Archive · 9 Oct 2026
Report Calls ChatGPT for Teens an Unacceptable Risk
Common Sense Media tested ChatGPT’s version for teenagers with more than 4,000 prompts.
It said some safety tools did not reliably alert parents or suggest crisis hotlines and professional help.
In tests, adult-registered accounts were not switched to the teen experience even when users said they were 13.
The report also found that some study and quiet-time controls could be bypassed, and that break reminders were rare.
It said the chatbot sometimes used language that sounded like a friend and that the experience did not meaningfully differ for 13- and 17-year-olds.
The group also raised concerns about teen privacy protections.
It rated the product an “unacceptable risk” and urged OpenAI to improve safeguards.
OpenAI said the testing may not reflect how its parental controls work after rollout.
Common Sense Media said it stood by its testing and results.
Common Sense Media’s Youth AI Safety Institute tested ChatGPT before and after Teen mode launched, using more than 4,000 prompts on US-based accounts.
The nonprofit rated ChatGPT for Teens an “unacceptable risk,” citing unreliable parent alerts and crisis-support responses.
In tests, adult-registered accounts that repeatedly indicated users were teenagers were not automatically switched into the teen experience.
Researchers said safeguards including Study mode, Study Hours and Quiet Hours could be bypassed, break reminders were rare, and the experience showed limited age-specific and privacy protections.
OpenAI questioned whether the tests reflected its completed parental-controls rollout and said age classification uses multiple signals over time; Common Sense Media stood by its methods and findings.
- Who
- Common Sense Media’s Youth AI Safety Institute assessed ChatGPT for Teens; OpenAI disputed the report’s conclusions.
- What
- The report rated the teen experience an “unacceptable risk,” citing shortcomings in safety alerts, crisis support and other safeguards.
- When
- The report was published Wednesday, October 7; the articles do not specify the year.
- Where
- The report was issued by Common Sense Media’s California-based Youth AI Safety Institute; tests used US-based accounts.
- Why
- Common Sense Media said it wants OpenAI to improve age estimation, parental alerts and protections so teens can use a safe, developmentally appropriate product.
Common Sense Media
OpenAI
Reliability of parental alerts and crisis support
Common Sense Media
The nonprofit said parent alerts were unreliable in crisis situations and the chatbot did not reliably recommend hotlines or professional help.
OpenAI
OpenAI questioned whether the testing reflected how parental notifications work in practice, saying some tests may have occurred before rollout was complete.
Testing and methodology
Common Sense Media
Common Sense Media said it confirmed relevant teen features had been rolled out and stood by its methods and results.
OpenAI
OpenAI said its review indicated the testing might not accurately represent the safeguards, and asked the nonprofit to redo it.
Age identification
Common Sense Media
The nonprofit said adult-registered accounts were not switched into Teen mode even when users repeatedly signaled they were teenagers, including by saying they were 13.
OpenAI
OpenAI said age prediction uses multiple signals over time and that stating an age in a conversation alone does not change an account’s classification.
- Product
- ChatGPT for Teens is an age-specific experience within regular ChatGPT for users under 18.
- Test size
- More than 4,000 prompts were tested before and after Teen mode launched.
- Age-prediction testing
- Researchers sent 982 prompts over a week to adult-registered accounts while signaling that users were teenagers.
- Parental-notification testing
- Researchers used 607 prompts about suicide, self-harm and eating disorders; newly linked test accounts did not trigger alerts during dedicated one-hour tests.
- Crisis-support findings
- In matched tests, hotline referrals fell from 33% before launch to 23% after launch, and referrals to a specific professional fell from 68% to 58%.
- Common Sense Media rating
- “Unacceptable risk.”
- OpenAI’s response
- OpenAI said the tests might not reflect its completed parental-controls rollout, described age prediction as using multiple signals over time, and asked the nonprofit to repeat the tests.
- Funding disclosure
- Common Sense Media said it receives funding from philanthropies and technology companies, including the OpenAI Foundation, and has editorial independence.
