OpenAI Restricts Teenagers’ Access to Weapons Information on ChatGPT

    OpenAI is imposing stricter limits on the practical information teenagers can obtain about weapons through ChatGPT. The new rules will apply even if a question appears harmless and there is no apparent intent to commit violence. The company said this is intended to prevent instructions about weapons and security systems that could be used in acts of violence.    OpenAI emphasized that its existing rules already prohibit ChatGPT from assisting with violent acts and directing teenagers to obtain weapons. The new approach will expand these protections further. At the same time, teenagers will still be able to use the service to prevent violence, de-escalate tensions, seek support, and contact a trusted adult.    The changes do not apply only to teenagers. OpenAI is also imposing stricter limits for all users on detailed reconstructions of past mass attacks. This includes extensive information about the weapons and equipment used in an attack, how they were obtained, and comparisons of the harm inflicted. The company believes such responses could lead to greater recognition of the attacker.    The new restrictions will work alongside the parental control features the company introduced last year. If a teenage profile linked to a parent’s account violates ChatGPT’s rules regarding violent threats or violence, the parent will be notified. The additional safeguards were announced after a wrongful death lawsuit was filed against the company last year. The company’s CEO, Sam Altman, also publicly apologized earlier this year for the failure to notify police about disturbing conversations related to the Tumbler Ridge shooting.    OpenAI plans to publish its first Community Safety Report in early 2027. Subsequent reports will be published every 6 months, and changes in the area of safety will also be shared with the public regularly. This plan was announced alongside the new restrictions on the information teenagers can obtain about weapons.