OpenAI says teens use ChatGPT for under 15 minutes a day as worries over risks grow

OpenAI revealed in its first report on teen usage that ChatGPT is engaged with by children for less than 15 minutes per day on average, with under 2% of teens interacting with the chatbot for over three continuous hours. The AI company's statement comes amid growing concerns over the potential risks that AI chatbots may pose to the mental health and developmental safety of adolescents.

These worries have been exasperated by research highlighting the adverse effects of social media on young users, prompting Meta to settle a lawsuit in August by agreeing to limit teen usage of Facebook and Instagram.

OpenAI reported that in nearly half of teen conversations that featured a break reminder, users either took a break or ended the chat within five minutes, suggesting that the company's efforts to regulate teen usage are effective. Users with access to the dedicated teen version of ChatGPT tend to send more education-related messages and utilize tools like learning visualizations and Study Mode.

The company is now developing a college application planning tool and new study features, such as flashcards and simplified quiz-making tools.

The teen version of ChatGPT, which was launched in August, automatically activates based on age information or when a user is estimated to be under 18, enabling parental controls and restricting conversations on sensitive topics. However, a report by non-profit Common Sense Media, based on more than a dozen accounts created for testing, revealed that ChatGPT for Teens frequently fails to notify parents promptly when teen accounts engage in conversations about suicide, self-harm, or eating disorders.

The study, which tested the chatbot's safeguards, found that the platform did not reliably suggest crisis hotlines or professional help. An OpenAI representative acknowledged that the testing does not accurately represent the functioning of the chatbot's teen safeguards, citing security and scale as reasons for the discrepancy. The company stated that it may take hours for parent and teen accounts to link and for safety alerts to activate, implying that the bulk of their testing may have begun and concluded before the activation of parental controls was complete.

Common Sense Media argued that while some of their test accounts were linked for extended periods, none of them produced timely alerts, with notifications only appearing for accounts that had weeks of sensitive-topic history.