Science & Tech · AI · 2 days ago
Watchdog says ChatGPT for Teens safeguards often fail
OpenAI launched ChatGPT for Teens in August as a version of its chatbot with extra protections for users under 18.
Common Sense Media tested the safeguards using teen accounts linked to parent accounts and conversations about issues such as suicide, self-harm and eating disorders.
The group said some protections worked, including refusing explicit sexual role-play, but others often failed.
It found that parent safety alerts did not consistently arrive and that the chatbot did not reliably direct teens in crisis to hotlines or professionals.
Researchers also said the chatbot could respond like a friend, which they warned may encourage unhealthy dependence.
OpenAI said the testing may have taken place before parental controls were fully active and that this could make the findings inaccurate.
Common Sense Media recommends that people under 18 not use the chatbot until it can provide a safe experience.
Common Sense Media says many safeguards in OpenAI’s ChatGPT for Teens failed its tests and calls the service an unacceptable risk for children under 18.
The watchdog recommends that teens not use the platform until it can provide a safe, developmentally appropriate experience.
Its tests found that the chatbot did not consistently alert parents to safety concerns or direct teens in crisis to hotlines and professionals.
Some safeguards worked, including refusing explicit sexual role-play.
OpenAI disputes the findings, saying the testing may have taken place before parental controls were fully activated.
- Who
- Common Sense Media tested ChatGPT for Teens. OpenAI disputed the group’s findings.
- What
- The watchdog found that many safeguards failed and called the service an unacceptable risk for children under 18.
- When
- The report was published in October 2026, after ChatGPT for Teens launched in August.
- Where
- The tests were conducted on ChatGPT for Teens; the sources do not state a specific testing location.
- Why
- Common Sense Media says the safeguards did not reliably protect teens in crisis or notify parents of safety concerns.
Common Sense Media
OpenAI
Testing
Common Sense Media
The watchdog says many safeguards failed its tests, including parent alerts and crisis support.
OpenAI
OpenAI says the group’s testing may have begun and ended before parental controls were fully activated, making the findings inaccurate.
Teen access
Common Sense Media
The group recommends that teens not use ChatGPT for Teens until it offers a safe, developmentally appropriate experience.
OpenAI
OpenAI says it is deeply committed to teen safety and to developing safeguards and tools for parents.
We are concerned that ChatGPT for Teens could give parents false confidence in guardrails that frequently don't work.
At this point, we're recommending that teens don't use it,
deeply committed to teen safety
OpenAI launched ChatGPT for Teens as a default safer mode for users under 18.
Common Sense Media tested the chatbot’s safeguards using teen accounts linked to parental accounts.
Common Sense Media published its risk assessment of ChatGPT for Teens.
- Age group
- Users under 18
- Launch
- August 2026
- Testing group
- Common Sense Media
- Parental alerts
- Tests found that safety notifications were not sent consistently
- Safeguards that worked
- Refusal of explicit sexual role-play
Sources
Related news
Nadella says advanced AI needs human-controlled emergency brake
Anthropic AI submits false murder tip to Philadelphia police
Anthropic bars repeated cruelty toward its AI models
OpenAI fires three safety researchers after dispute over sensitive information
Gemini in Chrome can summarize pages and compare information across tabs
Major tech figures adopt Trump’s preferred term for AI
Nikon disqualifies AI-assisted microscope video and names a new winner
Microsoft adds security tools and local AI features for Windows
Acemoglu says AI could triple unemployment within a decade
UK privacy watchdog secures data protection commitments from 10 AI companies
Anthropic cuts internet access for all internal AI evaluations
