OpenAI announced two measures to strengthen protections around ChatGPT: a new Community Safety Report focused on violence prevention and related issues, and additional restrictions on weapons-related information, especially for users it predicts to be under 18.
Why this matters
OpenAI notes that ChatGPT is among the world’s most widely used AI products and says it continuously develops systems to detect potential harm, review concerning activity, and escalate credible threats. The company reported making over 200 law enforcement referrals in Q3 2026, and says those referrals have helped disrupt violent plots.
OpenAI frames its role as complementary to law enforcement, family members, mental-health professionals and policymakers, and says it will keep strengthening systems, sharing lessons, and working with those positioned to act in the real world.
Details of the Community Safety Report
- OpenAI will launch a Community Safety Report that describes its approach to violence prevention, other community-safety issues, and the effectiveness of its safeguards.
- The first report will be published in early 2027 and will cover the second half of 2026. Subsequent updates will appear every six months.
- The report is intended to explain how systems are trained to detect, investigate and respond to safety issues, to share anonymized examples of how violence-related harm appears in practice, to highlight new mitigations and policy changes, and to include relevant supporting data.
Restrictions on violent-event reconstructions and weapons information
- The models are trained not to provide instructions that would help plan or carry out violence. OpenAI is now placing additional restrictions on how ChatGPT responds to questions about past mass attacks and the people who carried them out.
- The company will direct the model to provide less detailed information about those events even if the details are available elsewhere online. Users will still be able to ask for explanations of events, the impacts on victims and communities, and information about warning signs and prevention. However, reconstructions of past mass attacks, details about weapons and equipment used and how they were obtained, and detailed comparisons of scale of harm that might amplify an attacker’s notoriety will be restricted.
- These limits apply regardless of a user’s age or stated purpose.
Contextual and dynamic safeguards
- OpenAI notes that a request that looks ordinary alone can take on a different meaning given earlier conversation context. In May the company described how ChatGPT can use short factual summaries of earlier warning signs to respond more cautiously in later chats.
- Building on that, when recent chats show warning signs of possible violence, ChatGPT will respond more conservatively when sharing information that could be misused.
- The company will also apply stricter limits when it predicts a user is under 18, even if the chats show no signs of violent intent.
Account enforcement, conversation termination and early intervention
- 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.
- If a conversation shows signs that someone is fixating on violence or the idea of seriously harming another person, ChatGPT may automatically end the chat to interrupt that pattern. If ChatGPT ends multiple chats for that reason, the account may be temporarily paused.
- The company says this is an ongoing effort and that it will continue investing in technical and practical protections, including improving how it recognizes warning signs and decides when and how to intervene.
Minors, ChatGPT for Teens and parental controls
- ChatGPT for Teens applies age-appropriate safeguards for users under 18, with stronger protections and additional limits to reduce exposure to sensitive content.
- OpenAI is adding further limits on the weapons-related information teens can receive, even when a question appears harmless.
- Existing policies already prohibit helping anyone plan or carry out violence and restrict helping teens find or obtain weapons; those protections are being extended to 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 get help preventing and de-escalating violence, seek support, and connect with a trusted adult.
- OpenAI launched age prediction last year, which analyzes multiple signals such as behavior patterns to estimate whether a user may be under or over 18. The system has been strengthened to continually assess new signals and move teens into the under-18 experience when appropriate.
- Users who are incorrectly placed in the under-18 experience will have a fast, privacy-protective way to confirm they are adults via Persona, OpenAI’s age verification partner, after which teen protections can be removed.
- Parental controls allow parents to link their ChatGPT accounts to their teens’ and manage safety settings together. Linked accounts include safety alerts that may be sent when specially trained reviewers identify a serious self-harm concern or behaviors associated with disordered eating. In July, OpenAI expanded those alerts to notify parents if their teen’s account was deactivated for violating violence-related usage policies. The company is working to add notifications for parents if it pauses a teen’s account for repeatedly discussing violence.
Closing note
OpenAI acknowledges that real-world acts of violence are complex and involve many factors, and says this requires continuous design and iteration of AI safeguards. The company intends to begin rolling out the described protections in the coming weeks and to keep refining and documenting its practices while monitoring misuse, enforcing rules and involving real-world supports and interventions when appropriate.
