NYU Stern researcher @NateWitkin questions why cyber insurance rates keep falling if AI cyber risk is accelerating:

“Insurance rates for cyber risk declined by about 4% globally in Q2 of this year, and that’s actually the 12th consecutive quarter in which they’ve declined. This is very valuable signal that implies that at a minimum you shouldn’t overindex on the Hugging Face incident.”

“This is a plea for level headedness, but I think it would be helpful for safety folks to engage with these numbers just ’cause this is an avenue of criticism from folks like me and to an extent folks like Tyler.”

“Why are these numbers not moving? Is it because people are underestimating capabilities? Are they not taking the problem even as close to as seriously as they should or is it something else?”

Here is the link with video.  File under “Questions that are all too rarely asked.”

I am happy to admit that the answers here are far from obvious, and that I am myself expecting prices to rise somewhat.

I will continue to note that there are a remarkable number of ways, seen among other places on Twitter, to rephrase and to rationalize the statement: “I have the most remarkable and important and true macro risk story in the world to tell you.  Unfortunately, it does not correlate with any observed asset market prices.”