Our updated Preparedness Framework

In three linesOpenAI releases an updated Preparedness Framework for measuring and mitigating severe risks from frontier AI capabilities. The framework outlines evaluation methods for dangerous capabilities and safety safeguards.

Summary generated by Claude — human-verified