This is a really interesting perspective. I especially agree with the point that LLM-based security shouldn’t become a replacement for solid engineering practices. AI can make vulnerability discovery much faster, but that also means attackers get the same advantage.
The part about continuous fuzzing and property-based testing stood out to me. If the goal is to stay ahead of increasingly capable models, building systems that are continuously tested and difficult to break seems much more sustainable than simply relying on whichever model is strongest at the moment.
This is a really interesting perspective. I especially agree with the point that LLM-based security shouldn’t become a replacement for solid engineering practices. AI can make vulnerability discovery much faster, but that also means attackers get the same advantage.
The part about continuous fuzzing and property-based testing stood out to me. If the goal is to stay ahead of increasingly capable models, building systems that are continuously tested and difficult to break seems much more sustainable than simply relying on whichever model is strongest at the moment.