On October 30, 2026, OpenAI announced two community safety measures: a Community Safety Report, and tighter limits on the weapons-related information teenage users can receive. The first report is due in early 2027, covering the second half of 2026, with updates every six months thereafter; the new measures will begin rolling out in the next few weeks.
Details of past attacks will be scarcer for every user
The new rules are not only about teenagers. When ChatGPT answers questions about past mass attacks and the people who carried them out, it will provide less detail, even if that information is available elsewhere online. Restricted content includes reconstructions of attacks, details about the weapons and equipment used and how they were obtained, and detailed comparisons of the scale of harm that could amplify an attacker's notoriety. Users can still learn about the events themselves, how victims and communities were affected, how to recognize warning signs, and how to prevent violence; the baseline rule stands: no instructions that would help someone plan or carry out violence.
A conversation showing signs of violence may be cut short
In May, OpenAI described an approach that uses short, factual summaries of warning signs from earlier conversations so the model responds more carefully later. This announcement builds on that work: when recent chats show warning signs of possible violence, ChatGPT will respond more conservatively when sharing information that could be misused; even if a conversation shows no violent intent, stricter limits will apply when the system estimates the user is under 18. When a conversation shows signs of fixation on violence or seriously harming another person, ChatGPT will be able to end the chat automatically; after multiple chats are ended for that reason, the account may be temporarily paused. OpenAI said it made more than 200 referrals to law enforcement in the third quarter of 2026 involving imminent threats of violence.
For teens: even seemingly harmless questions will be restricted
ChatGPT for Teens, for users under 18, already applies age-appropriate safeguards, and this round tightens them further: practical information about weapons and security systems that teens can receive will be restricted, even when a single question appears harmless and shows no violent intent. Existing rules already prohibit helping teens find or obtain weapons; the new rules extend that to practical information that could be misused to facilitate violence. Teens can still use ChatGPT to learn how to prevent and de-escalate violence, seek support, and connect with a trusted adult. Age is judged by a strengthened age prediction system, which continually analyzes signals such as behavior patterns to estimate whether a user is under 18; adults who are incorrectly placed can confirm they are adults through Persona, the verification partner, and have the additional protections removed. Linked parental-control accounts receive safety alerts, which since July have included notifications when a teen's account is deactivated for violating violence-related usage policies. This site previously covered the risks reported for the teen version of ChatGPT (/article/2376).
The limits are clear too: automatically ending chats and pausing accounts are still being developed, and the safety report will not have its first edition until early 2027, when effectiveness data will be available. For parents, children will receive fewer answers to weapons-related questions; but age prediction can still make mistakes, and adults mistakenly placed in the under-18 experience will need to complete the verification process themselves to remove it.
