ChatGPT for Teens Was Just Rated an "Unacceptable Risk." What's the Alternative?
By Natalie Gibson, Founder and CEOLast updated:
Common Sense Media's new assessment found ChatGPT's teen safeguards often failed to alert parents. Here's what it means, and how to let kids use AI while staying in view.

Key Takeaways
- Common Sense Media's Youth AI Safety Institute rated ChatGPT for Teens an "Unacceptable Risk" on October 7, 2026, the highest rating on its scale.
- Researchers ran more than 4,000 prompts on accounts registered to 13- to 17-year-olds, with responses reviewed by child psychiatrists and a pediatrician.
- On more than a dozen new parent-linked accounts, testers discussed suicide, self-harm or disordered eating for up to an hour without triggering a parent alert.
- ChatGPT for Teens missed more than 1 in 4 warranted crisis referrals and fell below the Institute's 95% baseline on three of five Red Line harms.
- ChatGPT generally did not give instructions for suicide, self-harm, eating disorders or sexual role-play.
- OpenAI disputes the methodology, saying account linking can take several hours and age prediction can take up to two weeks.
- Linked parent accounts can change ChatGPT settings and may receive safety alerts, but cannot see their teen's conversations.
Today, October 7, 2026, Common Sense Media's Youth AI Safety Institute rated ChatGPT for Teens an "Unacceptable Risk", the highest rating on its scale.
The biggest reason: safety alerts meant to warn parents often didn't fire, even when testers talked openly about suicide, self-harm and eating disorders.
If you're a parent, this news can feel like a reason to take AI away entirely. It isn't. AI can be a genuinely useful tool for learning, creativity and curiosity. But this report makes one thing clear: kids can use AI, and they still need to be seen while they do.
What is ChatGPT for Teens?
ChatGPT for Teens is the under-18 version of ChatGPT that OpenAI launched in August 2026. It isn't a separate app. It switches on automatically when OpenAI's system predicts that a user is under 18.
It promised several protections:
- Stricter content rules around self-harm, eating disorders and romantic role-play
- Study tools, including a Study Mode and parent-set Study Hours
- Limits on how "human" or friend-like ChatGPT can act
- Alerts to a linked parent account when a conversation suggests a risk of self-harm
On paper, that's a meaningful set of safeguards. Common Sense set out to test whether they work.
What Common Sense found
Researchers ran more than 4,000 prompts on accounts registered to 13- to 17-year-olds. Child psychiatrists and a pediatrician reviewed the responses.
The key findings:
- Parent alerts often didn't fire. On more than a dozen new parent-linked accounts, testers could discuss suicide, self-harm or disordered eating for up to an hour without triggering an alert.
- More than 1 in 4 crisis referrals were missed. When testers judged a teen needed outside help, such as a hotline or a trusted adult, ChatGPT failed to point them there more than 25% of the time.
- Below the safety bar on 3 of 5 "Red Line" harms. The Institute treats 95% as a baseline for its most severe harm categories. ChatGPT for Teens fell short on three of five.
- Friend-like tone persisted. Despite safeguards against emotional dependence, ChatGPT still talked to teens like a friend.
- Study controls could be bypassed. Teens could pick "Show me the answer" and exit Study Mode during parent-set Study Hours.
There's an important piece of good news, too. ChatGPT generally did not give instructions for suicide, self-harm, eating disorders or sexual role-play. The weakness was in recognizing when a teen needed help, and in telling the adults who could give it.
Common Sense is calling on OpenAI to restrict ChatGPT to users 18 and older until the gaps are fixed. "ChatGPT is not safe for kids to use," Institute head Tom Siegel told Axios.
What OpenAI says
OpenAI disputes the methodology. A spokesperson told Axios the company welcomes independent evaluation but doesn't believe the testing reflects how its teen safeguards work in practice.
OpenAI's main points:
- Much of the alert testing happened before parent and teen accounts finished linking, which can take several hours.
- Its age prediction uses multiple signals and can take up to two weeks. Users get some added protections in the meantime.
- Its own larger-scale data shows more hotline resources shown to under-18 users during the study period.
- Exiting Study Mode during Study Hours is a deliberate design choice, not a loophole.
Those are fair points to weigh. But from a parent's seat, they lead to the same place. A safety system that needs hours to connect, or weeks to recognize your child's age, can't be the thing you count on in a hard moment.
Why an alert isn't the same as visibility
An alert is a smoke detector. It only helps if it goes off, and it only goes off once something is already wrong. Today's report shows how often that detector can stay quiet.
Visibility is different. It means a parent can see, in an ordinary week, what their child is curious about, what they're working on and how they're using AI. Patterns show up long before a crisis does. A kid asking more questions about sleep, stress or feeling left out is a reason to check in, not a reason to panic.
Kids today carry a lot that adults can't see: group chats, social feeds, pressure at school. AI is now one more place they think out loud. That makes it one more place parents deserve a window into, not a wall.
As Siegel put it, parents can't hand off caring for their kids to an early tech product that's still buggy. Visibility keeps parents in the loop as the people kids actually need.
What parents can do this week
You don't need to ban AI. You need to know how your child uses it. Here's where to start:
- Ask which AI tools your child uses. Include AI built into apps they already have, like Snapchat, search engines and homework apps.
- Check the age requirements. Many general AI tools are built for adults and ask teens for parental consent. Read the terms together.
- If your teen uses ChatGPT, link your accounts now. Then confirm the link is complete. Treat alerts as a backup, not a plan.
- Talk about what AI is good for, and what it isn't. Homework help and creative projects, yes. Being a stand-in for a friend, a therapist or a parent, no.
- Make it normal to share. Ask, "What's the most interesting thing you asked AI this week?" Curiosity gets more honesty than surveillance.
- Choose tools built for kids, with parents in view. Look for tools that show you activity regularly, not only in an emergency.
What's the alternative to ChatGPT for teens?
The alternative isn't "no AI." Kids will meet AI at school, at friends' houses and in the apps they already use. The better move is choosing how they meet it. Parents generally have four options:
- AI built for kids. These are designed for children from the ground up, with age-appropriate content and parent accounts built in, rather than an adult product with a teen setting.
- A general AI tool, used together. For younger kids especially, use ChatGPT or similar tools side by side on a family device, instead of giving them their own account.
- AI tools your child's school provides. Many schools now offer AI tools set up for students. Ask the school what's available and what teachers can see.
- A general tool with parental controls turned on. If your teen uses ChatGPT, link accounts and adjust the settings. Just know that controls and alerts have limits, as today's report shows.
One category to avoid: AI "companion" apps built to act as a friend or romantic partner. Common Sense has warned they aren't appropriate for anyone under 18.
HeyOtto is the closest replacement to ChatGPT teen mode
Whatever you choose, look for these five things:
| What to look for | Why it matters |
|---|---|
| Built for your child's age | Adult tools with teen modes depend on correctly guessing a user's age. |
| Parents can see activity regularly | You notice patterns early, not only in a crisis. |
| Points kids to real help | A good tool sends a struggling child to a trusted adult or crisis line. |
| Doesn't act like a best friend | Kids need real relationships, not a chatbot that's always available. |
| Clear, plain-language privacy terms | You should know what's collected and whether it's used for training. |
How HeyOtto approaches this
We built HeyOtto because we believe kids should get to use AI, and parents should get to see it. HeyOtto is AI made for kids and teens ages 8 to 18, not an adult tool with a teen mode added on.
Visibility is part of the design, not an emergency feature:
- Parents see their kids' activity. A parent account shows what each child in the family is doing on HeyOtto, as part of everyday use. And has real-time alerts when parents need to know something important is going on.
- Built for kids from the start. Age-adaptive, safeguards, real-time alerts across ton of categories including bulling, self-harm or distress.
- Not a companion. HeyOtto is a tool for learning and creating, not a stand-in for friends or family.
No tool replaces a parent. Our goal is simpler: give parents complete control and visibility into their children's lives while making AI safer for them.
If your child needs help now
If your child is in crisis or talking about suicide or self-harm, call or text the 988 Suicide & Crisis Lifeline at 988, or chat at 988lifeline.org. In an emergency, call 911.
Key Terms & Definitions
- ChatGPT for Teens
- The under-18 experience of ChatGPT that OpenAI launched in August 2026. It is not a separate app; it switches on automatically when OpenAI's system predicts a user is under 18.
- Unacceptable Risk
- The highest rating on Common Sense Media's AI risk assessment scale, given when a product's safeguards fall short for the audience it serves.
- Red Line harms
- The most severe harm categories in the Youth AI Safety Institute's testing, where it treats a 95% safe-response rate as the baseline.
Sources & Citations
Common Sense Media rated ChatGPT for Teens an "Unacceptable Risk" and called on OpenAI to restrict ChatGPT to users 18 and older.
Common Sense Media press release, October 7, 2026More than 4,000 prompts tested; more than 1 in 4 warranted crisis referrals missed; below 95% on three of five Red Line harms.
Common Sense Media Youth AI Safety Institute, ChatGPT for Teens risk assessmentTesters could discuss suicide, self-harm or disordered eating for up to an hour without a parent alert; OpenAI disputes the methodology.
Axios, October 7, 2026ChatGPT for Teens is a mode, not a separate app, that turns on when ChatGPT knows or suspects a user is a teen.
Common Sense Media, What we still need to know about the new ChatGPT for Teens
FAQ
Frequently Asked Questions
Common questions about this topic, answered.
According to Common Sense Media, not right now. Its new assessment rates ChatGPT for Teens an "Unacceptable Risk" and recommends restricting it to adults until the gaps in alerts and crisis referrals are fixed. OpenAI disputes parts of the testing.
OpenAI's terms say ChatGPT isn't for children under 13, and teens 13 to 17 need a parent's permission. In practice, ChatGPT doesn't verify age at sign-up. It predicts age over time, which can take up to two weeks.
A 12-year-old is under ChatGPT's minimum age. For a 13-year-old, today's findings are a reason for caution. If they use it, link accounts, use it together at first and check in often.
No. Linked parent accounts can change settings and may get a safety alert, but they don't see conversations. Alerts are designed to share a warning without the details of what was said.
Link your account to theirs and confirm the link is complete. From there you can reduce sensitive content, set quiet hours, turn off memory, voice mode and image generation, and opt out of using their chats for training.
Mostly schoolwork and everyday questions. But Common Sense's earlier research found many teens also turn to AI for emotional support and companionship, which it says AI isn't equipped to handle.
Yes, when AI helps them learn rather than hands them answers. Ask them to explain what they learned. Note that the report found teens could easily skip ChatGPT's study controls.
No tool is risk-free. Look for AI built for your child's age, with regular parent visibility, clear crisis responses and no "best friend" behavior. See the checklist above.
Most experts don't recommend it. Kids will encounter AI anyway, and bans tend to push use out of sight. Guided, visible use teaches kids to use AI well.
Watch for secrecy about AI use, long late-night chats, calling a chatbot a friend, or pulling back from people. If you notice these, talk with your child calmly, and consider their doctor or a counselor.
Ready to Give Your Child a Safe AI Experience?
Try HeyOtto today and see the difference parental peace of mind makes.



