Key facts
- OpenAI will restrict practical information on weapons and security systems that could be misused.
- ChatGPT will automatically end chats where users appear to be fixating on violence.
- OpenAI strengthened its age prediction system for users under 18.
- Details on past mass attacks, including weapons used and how they were obtained, will be limited for all users.
OpenAI is implementing new safeguards to prevent real-world violence by restricting the weapons-related information ChatGPT provides to teens. The company announced on Friday that it will limit practical details on weapons and security systems that could be misused, even when there are no signs of violent intent. For all users, ChatGPT will be able to automatically end chats fixated on violence, and accounts may be temporarily paused after multiple such discussions. OpenAI has also enhanced its age prediction system to better identify users under 18 and plans to inform parents if their teen's account is paused for discussing violence. Additionally, information on past mass attacks, including the weapons used and how they were obtained, will be restricted for all users. These new measures are set to roll out in the coming weeks.