Your Brain Has an Attack Surface

About a year ago, I began transitioning from software engineering to AI safety research. I was drawn into this by a question that arose while building runtime security for software systems: how do you impose constraints on a system you can’t fully observe? In AI safety, this question is at the very core: if we can’t reliably control how AI systems communicate and coordinate with each other, we can’t impose any other security properties on them. Since then, I’ve completed the AGI Strategy course ...