Teens spend less than 15 minutes a day on average using ChatGPT, while fewer than 2% use the chatbot for more than three consecutive hours, according to new figures released by OpenAI as concerns grow over the impact of AI tools on children.
OpenAI published its first report on teen ChatGPT use on Wednesday as AI adoption among young people continues to increase and technology companies face growing scrutiny over the potential effects of chatbots on adolescent mental health and safety.
OpenAI said that in nearly half of conversations where a break reminder appeared, the teen either stopped chatting or took a break within five minutes. The company presented the figures as evidence that its measures for managing teen usage are having an effect.
The company also said teenagers with access to its dedicated teen experience sent more education-related messages and used features such as learning visualizations and Study Mode. OpenAI is also working on additional educational tools, including a college application planning feature that could help students manage requirements, deadlines and financial-aid steps, along with flashcards and easier ways to create quizzes.
OpenAI began rolling out ChatGPT for Teens in August with additional safeguards and parental controls. The experience can be activated based on age information or when OpenAI estimates that an account belongs to someone under 18. It also places additional limits on conversations involving sensitive subjects.
However, a new report from nonprofit Common Sense Media has raised concerns about how those safeguards work in practice. The organization said its testing found that ChatGPT for Teens did not consistently alert parents quickly when test accounts discussed suicide, self-harm or eating disorders.
Common Sense Media said it tested more than a dozen accounts and found that the chatbot did not reliably provide crisis hotline information or recommend professional assistance in those situations. It also reported that parent notifications were sometimes delayed or did not arrive.
OpenAI disputed the findings, saying the testing did not accurately represent how its teen safety systems operate. The company said that security and technical requirements can mean it takes hours for parent and teen accounts to become fully linked and for parental safety alerts to become active.
According to OpenAI, some of the testing may have started and ended before the parental controls had finished activating. Common Sense Media, however, said some of its test accounts had remained linked long enough for the safeguards to be active, yet it still did not observe timely alerts.
The debate comes as technology companies face increasing pressure over how AI products affect younger users. Concerns have also been intensified by research into social media’s impact on children, with Meta recently agreeing to major settlements and restrictions involving teen use of Facebook and Instagram.
OpenAI’s latest usage figures suggest that most teenagers are not spending hours continuously chatting with ChatGPT, but the disagreement over safety alerts highlights ongoing questions about how effectively AI companies can protect younger users when conversations involve serious mental health concerns.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
