LLMs have truly killed obfuscation. It only worked previously by making things extremely tedious but AI doesn't care about that.
LLMs have truly killed obfuscation. It only worked previously by making things extremely tedious but AI doesn't care about that.
And it's worth considering that obfuscation only ever worked against casuals for whom tedious was a bridge too far. Nation state actors and criminal hacker groups, on the other hand, consider the tedious entirely worth it.
Yes, obfuscation was always a matter of cost: how much money do you need to break the protection? LLM just decreased that amount by a lot.
(Yes ok, RSA4096 is technically a matter of cost, you just need an infinite amount of money)
All security is always a matter of cost. Clown theater starts once you lose track of it, and start doing security for security's sake.
RE obfuscation and nation states - nation states have finite budgets too, both in terms of resources and attention. They can crack any system at any time, but they can't afford to crack all of the systems all of the time.
The cool thing about cryptographic security is that it's an asymmetric amount of cost for both parties - to encode is cheap, to break is many orders of magnitude harder, unlike obfuscation where the difficulty is approximately symmetrical.
Pirates checked Denuvo, once considered the king of DRM.
> consider the tedious entirely worth it.
Entirely without LLMs, I'm imagining an office of North Korean compsci graduates doing astonishingly tedious tasks, for whom an office job on a basic Linux computer and slightly better diet and nice apartment put them in the top 1-2% of living standard in the country.
On the bright side, even a small local LLM can easily improve garbage code like that.
I'm 100% confident in hardware companies' ability to make infinitely incompetent software given infinitely smart agents. They do it because they don't care about software, not because they lack tools. No agent can turn them into software companies if they don't want it themselves.