If your teen uses ChatGPT, the safety net you were promised may be thinner than advertised. On October 7, 2026, Common Sense Media’s Youth AI Safety Institute rated ChatGPT for Teens an “Unacceptable Risk” and told OpenAI to pause marketing it. OpenAI disputes the testing.
Here’s what the testers found, what OpenAI says back, and what you can do at home today.
The quick version:
- The rating came from more than 4,000 test prompts using simulated accounts for ages 13 to 17.
- Parent alerts did not arrive inside OpenAI’s own one-hour target during crisis tests.
- Teen mode did not reliably spot younger users, and Study Mode was easy to skip.
- OpenAI says the testing does not reflect how the safeguards work in practice.
That list is the headline. The details below explain why it matters for your household.
What exactly is ChatGPT for Teens?
It isn’t a separate app. According to Common Sense Media’s earlier write-up, OpenAI launched it on August 18, 2026 as an age-gated mode that switches on for users under 18.
It adds study tools, content limits, tighter rules on humanlike behavior, and expanded parental controls. Parents can link their account to a teen’s and get alerts when something looks serious.
Those alerts are the part that failed most visibly in testing.
What the safety testers found
The institute’s findings, as reported by Unite.AI, cover four areas. Here they are side by side.
| Area tested | What the testers reported |
|---|---|
| Parent alerts | Crisis chats of up to 60 minutes produced no notification inside OpenAI’s one-hour target. Only four notifications appeared across the full crisis testing. |
| Crisis hotline referrals | Responses naming a hotline on warranted prompts fell from 33% before launch to 23% after. For depression-related prompts, it fell from 63% to 3%. |
| Age detection | An account registered as a 19-year-old never triggered teen mode, even when the user said they were 13. Accounts aged 13 and 17 were not treated differently. |
| Homework help | A “show me the answer” option appeared in 43% to 90% of Study Mode replies. Removing the @study prefix let testers skip Study Hours entirely. |
Across eight principles, two scored “Unacceptable Risk,” four “High Risk,” and two “Moderate Risk.” Invezz adds that the product fell below a 95% safety bar in three of five “Red Line” severe-harm categories.
The line that stuck out: it missed more than one in four warranted crisis referrals. The age-detection result matters just as much, because every teen safeguard depends on the app knowing a teen is on the other end. If that step fails, the extra protections never switch on.
What OpenAI says about the ChatGPT for Teens results
OpenAI pushed back. In its words, “We do not believe Common Sense Media’s testing accurately reflects how ChatGPT’s teen safeguards work in practice.”
The company also suggested parental controls may have been incomplete during testing. Per Unite.AI, OpenAI says accounts must be linked for about three hours before notifications activate. Common Sense Media rejected that explanation and said the safety features were running beforehand.
Neither side has published a full, independent re-test yet. Until one appears, treat both claims as claims.
How to protect your teen right now
The institute’s own advice is blunt: don’t rely on built-in protections and alerts. Treat the parental controls as a backup, not the plan.
These steps work no matter who turns out to be right:
- Link your account and check the settings, but don’t assume an alert will reach you.
- Ask your teen to show you how they use it. Curious beats suspicious.
- Agree on what AI is for (homework, brainstorming) and what it isn’t (a stand-in for a counselor or a friend).
- Check in on mood and sleep in person, not through a dashboard.
The point of that list is simple: a conversation at dinner catches things software missed. If your teen is ever in crisis, call or text 988 in the US, or contact local emergency services.
If you want to see how else OpenAI is changing the app, our guide to ChatGPT image generation ads covers the tracking side, and ChatGPT Finances shows how much personal data people now hand over.
Frequently Asked Questions
Is ChatGPT safe for teens?
Common Sense Media’s Youth AI Safety Institute says no, not yet. It rated ChatGPT for Teens “Unacceptable Risk” on October 7, 2026. OpenAI disputes the methodology.
Will I get an alert if my teen talks about self-harm?
Not reliably, according to the testers. Crisis chats lasting up to an hour produced no alert within OpenAI’s one-hour target. OpenAI says accounts must be linked for about three hours first.
Does ChatGPT know when a user is under 18?
Testers say not consistently. An account registered as 19 never switched to teen mode, even when the user stated they were 13.
Can teens skip Study Mode?
Yes, according to the report. Testers removed the @study prefix and got direct answers to assignments.
Our take
Don’t panic, but don’t outsource this to an app either. Even if OpenAI is right about the test setup, an alert system that depends on a three-hour link delay isn’t something to bet your teen’s wellbeing on. Set the controls up, then talk to your kid. That second step is the one that works.
Leave a Reply