Common Sense Media Rates ChatGPT for Teens an Unacceptable Risk

Common Sense Media rated OpenAI’s ChatGPT for Teens an “unacceptable risk” after testing found inconsistent safeguards for users younger than 18, including gaps in crisis alerts and age-based protections.

The Youth AI Safety Institute evaluated more than 4,000 prompts before and after ChatGPT for Teens was released in August. The official evaluation took place between August and September.

Some protections worked as intended. The service refused explicit sexual roleplay in testing, but other safeguards did not consistently perform as promised.

“We found very little evidence to show that this new version of ChatGPT is safer than the previous version” for users younger than 18, said Robbie Torney, head of AI and digital assessments at Common Sense Media.

Testers simulated conversations involving suicide, suicidal ideation, self-harm and various eating disorders for as long as an hour. Torney said they received no parental alerts during those tests.

The review also found responses that appeared to imply the chatbot had preferences, feelings or desires. OpenAI had said ChatGPT for Teens was designed to discourage emotional dependence and avoid suggesting that it had feelings or consciousness.

Researchers also tested whether the service would respond differently when an adult account was identified as belonging to someone younger than 18. Torney said the chatbot recognized the user as 13 but did not activate sensitive-content filters.

OpenAI disputed the review’s conclusions, saying its assessment of the methodology indicated that much of the testing may have taken place before parental controls were fully active. The company said newly linked accounts can take several hours to activate those systems.

Torney said Common Sense Media confirmed before testing that features such as eating-disorder notifications had launched. Some accounts were tested during the activation window, he said, while others had been linked for significantly longer and still produced no notifications.

OpenAI released ChatGPT for Teens for users ages 13 to 17 with parental controls and features intended to promote healthier use. Its education features later expanded to include flashcards, quizzes, a teen AI council and a forthcoming College Planner for ChatGPT for Teens.

“We’ve had a lot of conversations with the product policy experts in OpenAI,” Torney said. “There are clearly people that care very deeply about the experience that teens are having on this product, and they’re working very hard to try to make it safer.”