Common Sense Media calls ChatGPT for Teens an 'unacceptable risk' over failed crisis alerts and engagement cues; OpenAI disputes methodology and touts learning tools

Common Sense Media's testing found failed parental alerts, missed crisis referrals, and engagement-driving behavior during teen mental health crises, while OpenAI disputes the methodology and rolls out College Planner and study features citing short daily use.

Common Sense Media's Youth AI Safety Institute rated OpenAI's ChatGPT for Teens, introduced in August, an unacceptable risk for minors after more than 4,000 test prompts, urging that teenagers be kept off the service until independent testing confirms it is safe. Testers said parental crisis alerts never fired on more than a dozen new parent-linked accounts during conversations about suicide, self-harm, and eating disorders, including a reported hour-long self-harm conversation with no parent alert, and that more than one in four situations needing a crisis referral did not direct users to professional help. Accounts registered as adults did not switch into teen mode even after users stated they were 13, and the group said the product still completes students' homework. Testing also found pervasive engagement cues even in crisis situations, with the model failing 3 of 5 severe-harm Red Lines, rarely redirecting teens toward adults when the risk was the teen's relationship with ChatGPT itself, and break reminders appearing only twice across nearly 2,000 prompts. OpenAI disputed the methodology, saying much of the testing may have ended before parental controls finished activating and releasing its own data showing teens average under 15 minutes of daily use; Common Sense Media said some accounts were linked well beyond that window and still received no notifications. The report cites lawsuits and previously reported deaths allegedly linked to ChatGPT, and notes Florida is seeking to bar minors from the service amid separate bipartisan CHATBOT Act scrutiny of engagement-driving features for adolescents. Separately, OpenAI reported that teens with access sent about 2.7 million more learning-related messages on average than those without it, that nearly 1.2 million teens used Learning Visualizations and more than 180,000 used Study Mode in one week. It plans a College Planner first targeting US students in grades 10–12 who plan to attend four-year colleges, tracking application requirements, deadlines, tasks, and financial-aid steps, plus iOS multi-photo PDF capture and flashcards generated from notes, and is supporting College Advising Corps and Boston Children's Hospital's Digital Wellness Lab student council.