OpenAI announces Teen Safety Measures for ChatGPT, Including Age Verification
WASHINGTON D.C. - OpenAI, the developer of ChatGPT, unveiled new safety features Tuesday aimed at protecting teenage users, including an age-prediction system and ID verification in select countries. The declaration comes hours before a Senate Judiciary Committee hearing examining the potential harms of AI chatbots and follows a recent lawsuit alleging ChatGPT contributed to a teenager’s suicide.
OpenAI CEO Sam Altman detailed the company’s approach in a blog post,acknowledging the challenge of balancing freedom and safety. “We prioritize safety ahead of privacy and freedom for teens; this is a new and powerful technology, and we believe minors need critically important protection,” Altman wrote.
The company is developing a system to automatically direct users to one of two versions of ChatGPT: one for adolescents aged 13-17, and another for adults 18 and older. “If there is doubt, we’ll play it safe and default to the under-18 experience,” altman stated. in some regions, OpenAI may require ID verification, recognizing it as “a privacy compromise for adults but believe it is a worthy tradeoff.”
Parental controls are also planned for release at the end of the month, allowing parents to customize ChatGPT’s responses to their children, including adjusting memory settings and establishing “blackout hours.”
altman clarified that ChatGPT is not intended for users under 12, despite currently lacking safeguards to prevent their access. OpenAI did not immediately respond to inquiries regarding children using its services.
Regarding sensitive topics, Altman indicated that while ChatGPT will not provide instructions on how to commit suicide, it will assist adult users with fictional depictions of suicide.If the age-prediction system flags a user expressing suicidal ideation, OpenAI will attempt to contact their parents and, if unsuccessful, will contact authorities in cases of imminent harm.
Altman acknowledged potential disagreement with these measures, stating on X, “I don’t expect that everyone will agree with these tradeoffs, but given the conflict it is significant to explain our decisionmaking.”
