I've been wondering - Is there any possibility that Nation States had AI models busy finding this stuff several years before we see things like Mythos or Opus showing up? Or are those models not only consumer but also cutting edge?
Nation State actors would have been using fuzzing tools and reverse engineering for years to find useful zero-days.
AI models likely don't give them anything new, rather it accelerates their current process a lot. But that's not to say that an AI may not discover some novel new technique.
Dirty Frag variant. Mitigations against that also work here.
Any more context here?
https://github.com/v12-security/pocs/tree/main/fragnesia
It seems addressed by this patch over here:
https://lists.openwall.net/netdev/2026/05/13/79
I know this AI stuff is annoying at times, but I do think it really improves Linux security.
I've been wondering - Is there any possibility that Nation States had AI models busy finding this stuff several years before we see things like Mythos or Opus showing up? Or are those models not only consumer but also cutting edge?
Nation State actors would have been using fuzzing tools and reverse engineering for years to find useful zero-days.
AI models likely don't give them anything new, rather it accelerates their current process a lot. But that's not to say that an AI may not discover some novel new technique.