OpenAI updated its preparedness framework, its first major update since the policy was originally released in December 2023.

Analysis of the update, comparing the new and old policies, will be added to the page soon.

A document of the new policy (nearly a complete rewrite) can be found at this link.