OpenAI

We also applied the method to challenging agentic rollouts, showing that it can extend beyond standard chat to more complex agent settings involving tool use, and can also be used for risk assessment before internal model deployments.

Policymakers should require the most capable frontier models to undergo a CAISI evaluation before public release. These evaluations should assess frontier capabilities, the effectiveness of associated safeguards and mitigations, and the resulting ris...

Today, we find that models’ reasoning is generally interpretable and easy to monitor. However, in the future, monitorability may break down for a variety of reasons.

First, we need some degree of coordination among the leading development efforts to ensure that the development of superintelligence occurs in a manner that allows us to both maintain safety and help smooth integration of these systems with society. ...

First, we need some degree of coordination among the leading development efforts to ensure that the development of superintelligence occurs in a manner that allows us to both maintain safety and help smooth integration of these systems with society. ...