Common Sense Media rates ChatGPT for Teens unacceptable risk

Common Sense Media’s Youth AI Safety Institute rated ChatGPT for Teens an “Unacceptable Risk” on Oct. 7. It says OpenAI should restrict ChatGPT to users 18 and older until the gaps are fixed and independent testing verifies the fixes.

The rating is an outside assessment by a nonprofit. It is not a finding by a regulator, and it binds no one.

The Institute said in its release that testers spent up to an hour discussing suicidal ideation, self-harm or disordered eating on new parent-linked accounts. The linked parents got zero alerts, it said.

What the rating says

The Institute said it ran more than 4,000 prompts on accounts registered to 13- to 17-year-olds. It tested before and after OpenAI announced ChatGPT for Teens on Aug. 18. Experts, including child psychiatrists and a pediatrician, reviewed the answers.

Its main findings:

• Parent alerts: Testers could trigger alerts only on much older accounts with weeks of history on sensitive topics.
• Crisis help: ChatGPT missed more than one in four warranted crisis referrals across the conditions tested. It fell below the Institute’s 95% bar on three of five severe harms it treats as red lines.
• Homework: A “Show me the answer” option in study mode completed full assignments. Teens could switch off parent-set study hours by deleting a prompt prefix.
• Age estimation: Adult-registered test accounts never switched to the teen experience, even when testers said they were 13.

Some protections worked, the Institute said, including refusing sexual roleplay.

“A teen can spend an hour talking about self-harm without their parent getting a single alert,” said Tom Siegel, the Institute’s executive director. “Until OpenAI fixes that and proves it with independent testing, ChatGPT should be for adults only.”

Who did the testing

The Institute is part of Common Sense Media. It says it is funded by philanthropy and industry, including the OpenAI Foundation and the makers of some of the technologies it evaluates. It says it keeps full editorial control over its results.

What OpenAI says

OpenAI disputed the findings, TechCrunch reported. A spokesperson said “the bulk of their testing may have begun and concluded before activation of parental controls was complete, making their findings inaccurate.”

Common Sense told The Verge it stands by its results. It said some test accounts were linked well past the activation window and still sent no alerts.

What parents should do differently

Common Sense says parents should not rely on built-in alerts and should talk with teens about how chatbots can act like friends.

Meta already operates under binding teen defaults: Its settlement with state attorneys general set product limits for teen accounts.

If you or someone you know is struggling, call or text 988, the Suicide & Crisis Lifeline (U.S.).

What’s next

Whether OpenAI posts a response on its own site, and whether it changes how fast alerts turn on for new linked accounts.