AI x Cyber Reading Group

​In this week's session, I will present some recent research I did in collaboration with Sean and Jack of Lyptus Research, a brand new AI safety research org from Australia!

​We applied METR's time-horizon methodology to offensive cybersecurity, grounded in a human expert study with 10 professional security practitioners.

In this reading group session, I'll present our motivation, methodology, results and interpretation of the work. Then, we'll continue our discussion from last week on cyber capabilities, benchmarks, and the concept of skill floors and skill ceilings in cyber security.

Full research writeup: https://lnkd.in/d5vBNGqR