[antirez]

Why this is the wrong analogy: finding hash collisions, while exponentially harder with N, is guaranteed to find, with enough work, some S so that H(S) satisfies N, so an asymmetry of resources used will have the side with more work eventually winning. But bugs are different: 1. different LLMs executions take different branches, but eventually the branches possible based on the code possible states are saturated. 2. if we imagine sampling the model for a bug in a given code M times, with M large, eventually the cap becomes not M (because of saturated state so of the code and the LLM sampler) but I, the model intelligence level. The OpenBSD SACK bug easily shows that: you can run an inferior model for an infinite number of times, it will never realize that the lack of validation of the start window if put together with the integer overflow then put together with the fact the branch where the node should never be NULL is entered produce the bug. So cyber security of tomorrow will not be like proof of work "more GPU wins", but better models and faster access to such models win.

[fhd2]

Agreed, it is different in terms of there being no guarantee that a specific piece of software even has an exploit. If you don't want to break into a specific piece of software, or even a specific system, I would argue that the law of averages applies: If you just invest enough, you'll likely find _something_ worth exploiting.

In other terms, I feel the argument from TFA generally checks out, just on a different level than "more GPU wins". It's one up: "More money wins". That's based on the premise that more capable models will be more expensive, and using more of it will increase the likelihood of finding an exploit, as well as the total cost. What these model providers pay for GPUs vs R&D, or what their profit margin is, I'd consider less central.

But then again, AI didn't change this, if you have more money you can find more exploits: Whether a model looks for them or a human.