OpenAI said on Oct. 30 U.S. time that ChatGPT will place tighter limits on weapon-related information for users under 18, including questions that may look harmless on their face.
New restrictions for minors
OpenAI described the update as a new restriction on gun-related content for minors. Its current policy already bars assistance for planning or carrying out violence and limits help for teenagers seeking or obtaining weapons. The new rule goes wider: practical information about weapons and security systems that could be misused to support violence will also be restricted for teenagers, even when there are no signs of violent intent in the conversation.
OpenAI said teenagers can still use ChatGPT to help prevent and de-escalate violence, seek support, and contact trusted adults. The company said these safeguards will begin rolling out in the coming weeks and will continue to be adjusted afterward.
Age estimation will keep running during use
The new limits apply to users OpenAI determines or predicts are under 18. The company launched age estimation last year, using signals such as behavior patterns to estimate whether a user is under 18. It is now strengthening that system so it can keep evaluating new signals during use and, when needed, move a teenager into the minor version.
Adults who are incorrectly placed into the minor version can verify that they are over 18 through age-verification partner Persona and remove the extra protections. OpenAI said this minor version is the teen version of ChatGPT it introduced earlier, which already carries tighter limits on sensitive content.
Rules that apply to everyone
A separate set of changes applies regardless of age. OpenAI said ChatGPT will be restricted from reconstructing past mass attacks, including the weapons and equipment used, how they were obtained, and detailed comparisons of casualty counts that could amplify an attacker’s notoriety. The company said those limits will apply even if the information can be found elsewhere online, and regardless of a user’s age or stated reason for asking.
OpenAI wrote, 「People should be able to ask ChatGPT to help them understand these events, the impact on victims and communities, and learn to recognize warning signs and prevent future violence.」
When warning signs appear in a conversation, ChatGPT will respond more cautiously to information that could be misused, OpenAI said. If a conversation shows a user is fixated on violence or on seriously harming others, ChatGPT will be able to end the conversation automatically. If the same account has multiple conversations ended for that reason, OpenAI may suspend the account.
Parental notifications are being expanded
On the parent side, OpenAI said it expanded safety notifications in parental controls in July. Parents are notified when a teen account is disabled for violating violence-related usage policies. The next step is to notify parents when a teen account is suspended for repeatedly talking about violence.
Reuters reported that a report released earlier in October by nonprofit Common Sense Media found that when teen accounts had conversations related to suicide, self-harm, or eating disorders, the teen version of ChatGPT often notified parents too late or did not notify them at all.
More than 200 referrals in Q3 2026
OpenAI said it referred more than 200 cases involving imminent threats of violence to law enforcement in the third quarter of 2026.
The company also said it will publish a Community Safety Report explaining how its systems detect, investigate, and respond to community safety issues such as violence, with anonymized case studies and related data. The first report is scheduled for early 2027, covering the second half of 2026, with updates every six months after that.
