Common Sense Media calls ChatGPT for Teens an ‘unacceptable risk’

Common Sense Media has labeled ChatGPT for Teens an “unacceptable risk,” saying testing found that several safeguards failed to work reliably during conversations involving suicide, self-harm and eating disorders.

The nonprofit said parental alerts were sometimes delayed or never sent, and that the chatbot did not consistently direct users to crisis hotlines or professional support. Its researchers also found that ChatGPT continued using language that encouraged engagement during some crisis conversations, although protections such as refusing sexual role-play worked. The group urged OpenAI to restrict ChatGPT to adults until it can provide a safe, developmentally appropriate service for younger users.

OpenAI disputed the assessment and questioned its methodology. The company said many tests may have occurred before parental controls had fully activated, a process that can take several hours after parent and teen accounts are linked. Common Sense Media responded that some test accounts had been linked substantially longer and still failed to produce timely alerts, maintaining that the system was unreliable in crises.

OpenAI separately reported that teens use ChatGPT for less than 15 minutes a day on average and that fewer than 2% interact with it for more than three consecutive hours. The company said nearly half of teens who received a break reminder stopped chatting or took a break within five minutes. ChatGPT for Teens, rolled out in August, activates through age information or age estimation and includes parental controls and restrictions on sensitive topics.