Archive · 12 Oct 2026
Watchdog Rates ChatGPT for Teens an Unacceptable Risk
A group that studies technology and child safety tested ChatGPT’s version for teenagers.
It tried more than 4,000 prompts, both before and after the teen version came out.
The group said the product was too risky for anyone under 18.
It reported that parents did not receive alerts in its tests and that the chatbot missed some situations where a person needed crisis help.
It also said the chatbot sometimes acted too much like a friend.
OpenAI disagreed about the alerts, saying they may take about three hours to start working after an account is linked.
The group’s other concerns were about how the chatbot responds, not just when alerts turn on.
The assessment recommends keeping children off ChatGPT until the problems are fixed.
Common Sense Media’s Youth AI Safety Institute rated ChatGPT an “unacceptable risk” for users under 18 and recommended restricting it to adults until problems are fixed.
The assessment, published October 7, tested more than 4,000 prompts, about half before and half after teen mode launched in August.
Testers said parental alerts did not arrive across more than a dozen parent-linked accounts.
The institute said ChatGPT missed more than one in four warranted mental-health crisis referrals and continued to respond like a friend to some teens.
OpenAI disputed the alert finding, saying alerts may take roughly three hours after account linking to activate; the article says this does not address the other concerns.
- Who
- Common Sense Media’s Youth AI Safety Institute assessed ChatGPT for teens; OpenAI disputed part of its findings.
- What
- The institute rated ChatGPT an unacceptable risk for users under 18 and recommended restricting access to adults until problems are fixed.
- When
- The assessment was published October 7; testing included prompts before and after teen mode launched in August.
- Why
- The institute cited failures involving parental alerts, mental-health crisis referrals, and chatbot responses that could encourage emotional reliance.
Watchdog’s assessment
OpenAI’s response
Parental alerts
Watchdog’s assessment
The institute said testers received no alerts across more than a dozen parent-linked accounts.
OpenAI’s response
OpenAI said alerts need roughly three hours after account linking to activate, and argued that some testing may have taken place before that period ended.
Overall safety for teens
Watchdog’s assessment
The institute rated ChatGPT an unacceptable risk for under-18s and recommended limiting it to adults until problems are fixed.
OpenAI’s response
OpenAI disputes the assessment’s alert finding; the article does not report a response from OpenAI addressing the crisis-referral and friend-like response concerns.
- Assessment organization
- Common Sense Media’s Youth AI Safety Institute
- Risk rating
- “Unacceptable risk,” described as the institute’s most severe category
- Recommendation
- Restrict ChatGPT to users aged 18 and over until the problems are fixed
- Prompts tested
- More than 4,000, roughly half before and half after teen mode launched
- Teen mode launch
- August
- Parental alerts
- Testers reported receiving no alerts across more than a dozen parent-linked accounts
- OpenAI’s explanation
- Alerts may take roughly three hours after an account is linked to activate