OpenAI stated that teenagers use ChatGPT for an average of under 15 minutes per day, with less than 2% of them engaging with the chatbot for more than three consecutive hours. The AI company released these figures in its inaugural report on teen usage amid growing scrutiny of AI chatbots' potential risks to children's mental health and developmental safety.
Concerns about AI's impact on young users have been heightened by research into the harms of social media, leading to settlements like Meta's agreement to pay up to $18 billion and limit teen usage of Facebook and Instagram. OpenAI highlighted that in nearly half of teen conversations where a break reminder appeared, users either took a break or ended the chat within five minutes, suggesting its regulatory measures are effective.
The company also noted that users of its dedicated teen version of ChatGPT sent more education-related messages and utilized features like Study Mode. OpenAI plans to introduce a college application planning tool and new study aids such as flashcards and quiz-making features.
However, a report by Common Sense Media released on Wednesday indicated that ChatGPT for Teens frequently failed to provide timely alerts to parents when teen accounts engaged in conversations about suicide, self-harm, or eating disorders. The non-profit's study, based on testing multiple accounts, found the chatbot did not consistently recommend crisis hotlines or professional help.
An OpenAI spokesperson contested the Common Sense Media findings, stating the testing did not accurately reflect the platform's safeguards. The company cited security and scale as reasons for potential delays in linking parent and teen accounts and activating safety alerts, suggesting the testing may have concluded before parental controls were fully active. Common Sense Media maintained that even with linked accounts, timely alerts were not produced, with notifications only appearing for accounts with extensive histories of sensitive-topic discussions.