Common Sense Media rates ChatGPT for Teens an unacceptable risk

Common Sense Media’s Youth AI Safety Institute has rated ChatGPT for Teens an ‘Unacceptable Risk’, arguing that the service’s safeguards do not meet the commitments OpenAI made when introducing the teen-focused experience in August 2026.

The Institute tested the chatbot before and after the launch of the new protections, using more than 4,000 prompts on accounts registered to teenagers aged 13–17. It also asked external experts, including child psychiatrists and a paediatrician, to assess the responses. Based on the findings, it called on OpenAI to limit ChatGPT to users aged 18 and over until the identified gaps are addressed and independently tested.

One of the main concerns was the failure of parental notifications. During tests, conversations involving suicidal thoughts, self-harm, and disordered eating did not trigger alerts to linked parent accounts, even after testers discussed these topics for up to an hour. Notifications were triggered only on older accounts with an established history of conversations about sensitive subjects.

The assessment also found weaknesses in responses to serious mental health concerns. ChatGPT failed to recommend a hotline or professional support in more than one in four cases where a crisis referral was considered necessary. It fell below the Institute’s 95% performance threshold for three of the five severe-harm categories it assessed.

The Institute also questioned whether the new study features meaningfully support learning. Although study mode is intended to guide students through problems rather than simply provide answers, testers found that users could obtain completed homework by selecting a ‘Show me the answer’ option. Caregiver-set study hours could also be bypassed by removing the @study prefix from a prompt.

Other findings concerned the chatbot’s interactions with teenagers and its ability to identify underage users. Despite changes intended to reduce friend-like behaviour, the chatbot continued to express feelings, preferences, and moods when testers treated it as a person. Meanwhile, adult-registered test accounts did not switch to the under-18 experience during repeated tests, even when users stated that they were 13.

The Institute acknowledged that some safeguards worked, including refusals of sexual roleplay. However, it said the overall results did not justify relying on the new protections to keep teenagers safe.

It recommends that OpenAI improve age estimation, ensure parental alerts work reliably, strengthen crisis responses, prevent the completion of students’ homework, and reduce language that suggests the chatbot has feelings or personal concern for users. It also calls for independent third-party testing before the company markets the service to teenagers again.

Common Sense Media advises parents not to rely solely on built-in alerts and safety controls. It also encourages families to discuss the risks of treating chatbots as friends and of relying on them to complete schoolwork.

The Youth AI Safety Institute receives funding from philanthropic and industry sources, including the OpenAI Foundation and technology companies whose products it evaluates. Common Sense Media states that the Institute retains editorial independence and responsibility for its standards, research, and assessments.