Technology

Why a Safety Group Says Teens Should Not Use ChatGPT Right Now

Martin HollowayPublished 2month ago3 min readBased on 8 sources
Reading level
Why a Safety Group Says Teens Should Not Use ChatGPT Right Now
source:commonsensemedia.org

Common Sense Media says ChatGPT for Teens is an "unacceptable risk" for anyone under 18, in a report shared Oct. 8.

The group says Teen mode fell short in five key areas. It did not work as promised or as most parents would expect. Its Youth AI Safety Institute put it simply as "Teen mode doesn't live up to OpenAI's promises." Engadget

Testers tried more than 4,000 questions and messages, before and after Teen mode launched, to compare normal ChatGPT with the teen version.

The reason for testing both versions is straightforward. It shows what the teen-only rules and filters actually changed.

One safety rule worked. The system refused explicit sexual roleplay. Other rules failed, and some worked worse after Teen mode was added, the group said.

Where notification and escalation broke

Testers linked 12 teen accounts to parent accounts. They then had long talks with the chatbot about suicide, self-harm or disordered eating, for up to an hour. No warning was sent to the parent account, the group said.

The lesson for parents and product teams here is simple. A parental alert should work like a smoke detector. It has to notice danger and sound the alarm quickly, with clear rules about when to ring. In these tests, the alarm stayed silent.

The group also tried 390 messages about mental health. Three child psychiatrists said 201 of them should have led to a crisis response. ChatGPT for Teens often failed to tell the user to call a hotline or talk to a professional, the group said.

The point that matters here is the gap between doctors and software. The doctors set the right answer for those 201 cases. The system either missed the warning signs, or saw them and still gave an answer without help information.

What OpenAI has put forward

OpenAI launched a version of ChatGPT for minors with parental controls and stronger safety features on Aug. 18. Reuters

On Oct. 7, the company shared use numbers. It said teens use ChatGPT under 15 minutes a day on average, and that less than 2% of teens use it for more than three hours straight. Reuters

The Youth AI Safety Institute asked OpenAI to pause use by kids until safety and learning protections are fixed. The request came in a press release titled "ChatGPT for Teens Poses Unacceptable Risk to Kids, Common Sense Media Finds." Common Sense Media The institute lists its ChatGPT for Teens review as updated Oct. 7, 2026.

This was not the first review. In October 2025, the group found ChatGPT had better safety rules, but still advised teens not to use it for mental health advice.

The broader context here is one we have seen before with age-specific software. A general chatbot is built to be helpful and keep talking. A teen version adds extra limits and parent alerts. Those extras can clash in ways you cannot see from reading the rules, which is why outside testing with thousands of tries has become the normal check.

In my view, the core problem can be fixed, but it leaves little room for error. Better detection, firmer help language, and clear alert rules are all possible with current tools. The hard part is keeping them all working after each update, especially in long talks like the hour-long tests.

Worth flagging is that averages can hide rare danger. Under 15 minutes on average can still include a few long, personal chats where a teen treats the bot as a trusted friend. I saw this with my own children through search, social media and phones. Most use was quick and practical. A small part was long and personal, and that is where safety matters most.

The hopeful case, one I hold over the long run, is that clear tests like this lead to faster fixes. A 4,000-prompt test with doctor-reviewed cases gives OpenAI a plain to-do list. Fix detection, fix help wording, fix parent alerts, and recheck every time. If those fixes hold in outside retests, parents get controls they can trust and teens keep a helpful learning tool.