Here’s roughly how I’m thinking about red-teaming CoT monitoring systems. I think it’s all self-explanatory, and I’d be interested in feedback on the threat model and types of evidence I think should be collected for the different nodes.
Here’s roughly how I’m thinking about red-teaming CoT monitoring systems. I think it’s all self-explanatory, and I’d be interested in feedback on the threat model and types of evidence I think should be collected for the different nodes.