,
3 tweets,
1 min read
Read on Twitter
There seem to be a bunch of projects where researchers solved the problem of “make a working software exploit that satisfies these specific constraints.” Seems like it would be more valuable to solve this problem more generally. wired.com/story/malware-…
That is, build a general tool that takes in an exploit and the specific constraints (e.g., resulting exploit must have this symbol mix, length, etc.) and just solves for an equivalent exploit using SMT or something.
This is a harder problem and also probably results in fewer research papers, but would be a whole lot more interesting from a research perspective.
