OpenAI Introduces New Violence Safeguards and Safety Report

OpenAI has introduced new safety measures to address imminent threats of violence. The company made over 200 law enforcement referrals in the third quarter of 2026 regarding such threats. ChatGPT models are now restricted from providing detailed information about weapons and equipment used in past mass attacks. These restrictions apply regardless of the user's age or reason for asking. The system will also respond more conservatively when recent chats show warning signs of possible violence.

For users predicted to be under 18, stricter limits on information sharing are applied. ChatGPT for Teens places further limits on weapons-related information. The platform can automatically end a chat if it detects signs that someone is fixating on violence or seriously harming another person. OpenAI may temporarily pause an account if multiple chats are ended due to signs of violent fixation. Parents can link their accounts to teens to manage safety settings and receive alerts if an account is deactivated for violating usage policies.

OpenAI plans to publish its first Community Safety Report in early 2027. This report will cover data and events from the second half of 2026. The company will update the report every six months after the initial publication. The new safety safeguards detailed in the report will begin rolling out in the next few weeks.