OpenAI discloses 200+ police referrals
OpenAI says it made more than 200 law enforcement referrals from ChatGPT in Q3 2026, and it is pairing a new semiannual safety report with tighter limits on violence and weapons content.
OpenAI announced that it made more than 200 law enforcement referrals involving imminent threats of violence in the third quarter of 2026. The company says mental health and behavioral experts help it assess difficult cases.
The company also committed to a semiannual Community Safety Report covering violence prevention and how well its safeguards work. The first edition is due in early 2027 and will cover the second half of 2026. OpenAI says the report will include anonymized examples, new mitigations and supporting data.
OpenAI is also changing how ChatGPT responds:
- Past mass attacks: ChatGPT will give less detail about past mass attacks and the people who carried them out, even when that information is available elsewhere online. That includes the weapons used, how they were obtained and comparisons of scale that could add to an attacker's notoriety. These limits apply to all users.
- Conversation history: When recent chats show warning signs of possible violence, ChatGPT will respond more conservatively. OpenAI is working to apply the same stricter limits to users it predicts are under 18.
- Ending chats: ChatGPT will be able to automatically end conversations that show fixation on seriously harming another person. Repeated endings may lead to a temporary account pause.
- Teens: The teen experience will restrict practical information about weapons and security systems, even when there is no sign of violent intent.
OpenAI says its age prediction system now keeps checking new signals and can move users into the under-18 experience as it goes. Adults placed there by mistake can verify their age through Persona.
OpenAI did not provide figures on how accurate these systems are.