OpenAI to publish a Community Safety Report and restrict weapons information for teens

OpenAI is putting its violence-prevention work on a publishing schedule. On October 30, the company said it will launch a Community Safety Report covering violence prevention and other community-safety issues, alongside tighter restrictions on firearms content for minors and on how ChatGPT answers questions about past mass attacks.

The announcement pairs transparency with enforcement. OpenAI said it made more than 200 law-enforcement referrals in the third quarter of 2026, bringing imminent threats of violence to the authorities’ attention, and that those referrals are making a positive difference in disrupting violent plots. The company said it continues to rely on mental-health and behavioral experts to assess difficult cases.

A report on violence prevention, twice a year#

The Community Safety Report is a commitment to explain, in public, how the safeguards work and how well they work. OpenAI said the report will explain how it trains its systems to detect, investigate and respond to community-safety issues, share anonymized examples of how violence-related harm manifests and what the company is learning, highlight new mitigations and policy changes, and include relevant data to support its findings.

The first report is due in early 2027 and will cover the second half of 2026, with updates every six months after that. The point, OpenAI said, is to give people a clear picture of its approach, the risks it sees in practice, and the steps it takes to address them. AI companies, it added, have a distinct role alongside law enforcement, family members, mental-health professionals and policymakers.

Tighter boundaries for every user#

OpenAI’s models are already trained not to give instructions that would help someone plan or carry out violence. The new step restricts how ChatGPT responds to questions about past mass attacks and the people who carried them out, directing the model to provide less detailed information even when that information is available from other online sources.

What is restricted, specifically, is reconstructions of past mass attacks — including details about the weapons and equipment used and how they were obtained — and detailed comparisons on scale of harm that OpenAI said could amplify an attacker’s notoriety. The limits apply regardless of a user’s age or reason for asking. People can still ask ChatGPT to help them understand these events, how victims and communities were impacted, and how to recognize warning signs and prevent future violence.

When earlier chats change the answer#

A request that looks ordinary on its own can take on a different meaning in light of an earlier conversation. In May, OpenAI shared how ChatGPT can use short, factual summaries of earlier warning signs to respond more carefully in a later chat. It is building on that work so that when recent chats show warning signs of possible violence, ChatGPT responds more conservatively when sharing information that could be misused. It is also working to apply those stricter limits when it predicts someone is under 18, even if their chats show no signs of violent intent.

OpenAI bans accounts it determines have violated its policies on violent activity. It is also working with mental-health and violence-prevention experts on additional, automated ways to intervene earlier: when a conversation shows signs that someone is fixating on violence or the idea of seriously harming another person, ChatGPT will be able to automatically end the chat to interrupt that pattern in the moment. If ChatGPT ends multiple chats for that reason, OpenAI may temporarily pause the account. The company described this as an ongoing effort, with work continuing on recognizing warning signs reliably and deciding when and how to intervene.

What changes for teens#

ChatGPT for Teens already applies age-appropriate safeguards for users under 18. OpenAI is now placing further limits on the weapons-related information teens can receive, even when a question appears harmless. Its policies already prohibit helping anyone plan or carry out violence and restrict helping teens find or obtain weapons; the extension restricts practical information about weapons and security systems that could be misused to facilitate violence, even when there is no sign of violent intent. Teens can still use ChatGPT to help prevent and de-escalate violence, seek support, and connect with a trusted adult.

Underpinning the teen experience is age prediction, which analyzes signals such as behavior patterns to estimate whether a user may be under or over 18. OpenAI said it has strengthened the system to continually assess new signals as people use ChatGPT and move teens into the under-18 experience when appropriate. Users incorrectly placed in the under-18 experience can confirm they are adults through Persona, OpenAI’s age-verification partner, and have the additional teen protections removed.

Parental controls let parents link their ChatGPT accounts to their teens’ accounts and manage safety settings together. Linked accounts include safety alerts that may be sent when specially trained reviewers identify a serious self-harm concern and behaviors that may be associated with disordered eating. In July, OpenAI expanded those alerts to notify parents if a teen’s account was deactivated for violating its usage policies on violence, and it is working to introduce notifications if a teen’s account is paused for repeatedly discussing violence.

What to watch#

The safeguards detailed in the announcement will begin rolling out in the next few weeks, and OpenAI said it will continue to refine them. The first real test of the transparency commitment arrives with the report itself in early 2027: whether its data and anonymized examples let outsiders judge the safeguards, rather than simply take OpenAI’s word for them.

OpenAI closed with a note of caution about the problem itself. Acts of real-world violence are heartbreaking and complex, it said, with many contributing factors — and a reminder that designing safeguards for them means accounting for the complicated realities of human behavior. The company said it will continue to invest in, iterate on and share what it learns about implementing the protections it is pursuing.

Sources: OpenAI’s October 30, 2026 announcement, here; Unite.AI’s October 30, 2026 report on the announcement, here. Cover and inline images are AI-generated editorial illustrations.