On Friday, Oct. 30, OpenAI said ChatGPT will be able to end a chat on its own when the conversation shows signs that someone is fixating on violence or on seriously harming another person. If ChatGPT ends several chats for that reason, OpenAI says it may pause the account for a while. When someone's recent chats show warning signs of possible violence, ChatGPT will also answer more cautiously when sharing information that could be misused. The company says the new safeguards will begin rolling out in the next few weeks.
Teens get tighter limits. OpenAI's rules already bar helping anyone plan or carry out violence and limit helping teens find or get weapons. ChatGPT will also hold back practical information about weapons and security systems that could be misused, even when a teen's question looks harmless and shows no sign of violent intent. Teens can still use ChatGPT to seek support, reach a trusted adult, or help prevent or calm violence. OpenAI says it is also working on alerts that tell parents who have linked their accounts to their teen's when that account is paused for repeatedly discussing violence.
OpenAI uses an age-prediction system, launched last year, that looks at signals such as behavior patterns to guess whether a user is under 18. OpenAI says that system now keeps weighing new signals as people use ChatGPT and moves people it judges to be under 18 into the teen version when it sees fit. Adults placed there by mistake can confirm their age through Persona, the company OpenAI uses to check ages, and have the extra limits removed.
For everyone, ChatGPT will give less detail about past mass attacks. It will hold back on reconstructing an attack and on the weapons and equipment used or how they were obtained, even when those details are online elsewhere. It will also avoid detailed comparisons of how much harm attacks did, which OpenAI says could add to an attacker's notoriety. These limits apply whatever a user's age or reason for asking. People can still ask how victims and communities were affected and how to spot warning signs.
OpenAI says it brings imminent threats of violence to law enforcement's attention, and that it made more than 200 referrals to law enforcement from July through September 2026. It plans to publish a report on violence prevention and other safety issues, and how well its safeguards work. The first is due in early 2027 and will cover the second half of 2026, with new reports every six months after that. The company calls its violence safeguards an ongoing effort and says it is still working out how to spot warning signs reliably and when to step in. It has not said how often ChatGPT might end a chat by mistake, or how long a paused account stays paused.