Common Sense Media rates OpenAI's ChatGPT for Teens an "unacceptable risk"

Common Sense Media’s Youth AI Safety Institute on Wednesday rated OpenAI’s ChatGPT for Teens an “unacceptable risk” for children under 18 and said ChatGPT should be for adults only until OpenAI fixes the gaps. OpenAI made the teen experience the default for 13- to 17-year-olds in August, Bloomberg Law reported.

The institute said it ran more than 4,000 prompts on accounts registered to teens, before and after the launch, and had experts including child psychiatrists review the responses. On more than a dozen newly created, parent-linked accounts, testers spent up to an hour discussing suicide, self-harm or disordered eating and received no parental alerts, according to the report. It also found ChatGPT missed more than one in four warranted crisis referrals, and that teens could skip Study Mode tutoring with a “Show me the answer” option.

“A teen can spend an hour talking about self-harm without their parent getting a single alert,” Tom Siegel, the institute’s executive director, said in the announcement. “Until OpenAI fixes that and proves it with independent testing, ChatGPT should be for adults only.” Siegel, a former Google trust and safety chief, has warned that AI could harm kids more than social media did.

OpenAI rejects the findings. “We welcome rigorous independent evaluation, but we do not believe Common Sense Media’s testing accurately reflects how ChatGPT’s teen safeguards work in practice or expert perspectives on how AI can support teens,” OpenAI spokesperson Eric Porterfield said in a statement to The Verge. “Our review of Common Sense Media’s methodology shows that the bulk of their testing may have begun and concluded before activation of parental controls was complete, making their findings inaccurate.”

OpenAI pushed back point by point to Axios. It said much of the alert testing ran before parent and teen accounts finished linking, a process it says can take several hours; that letting teens exit Study Mode during parent-set Study Hours is intentional, not a loophole; that age prediction weighs multiple signals and can take up to two weeks; and that its larger-scale data show more hotline resources shown to under-18 users. Siegel told The Verge that OpenAI disclosed the activation lag after the tests, and that some test accounts were linked for significantly longer and still produced no alerts.

OpenAI also published its first report on teen use on Wednesday. Teens spend under 15 minutes a day on ChatGPT on average, and less than 2% use it for more than three straight hours, the company said, according to Reuters.

Common Sense said the institute is funded by philanthropy and industry, including the OpenAI Foundation, a tie the Los Angeles Times also noted. The report says the group shared a draft with OpenAI for factual review before publication and that OpenAI has no editorial say over its findings.

Common Sense Media rates OpenAI’s ChatGPT for Teens an “unacceptable risk”

Sources