A widely discussed potential risk from LLMs is increased access to biothreat creation information. Building on our Preparedness Framework, we wanted to design evaluations of how real this information access risk is today and how we could monitor it going forward.

→ View original post on X — @openai,