AI models can generate exploit code at lightning speed (www.theregister.com)
from cm0002@lemmy.world to cybersecurity@infosec.pub on 22 Apr 02:34
https://lemmy.world/post/28544135

#cybersecurity

threaded - newest

otter@lemmy.dbzer0.com on 22 Apr 04:09 next collapse

Uh. Duh?

noctivius@lemm.ee on 22 Apr 07:48 next collapse

When it has a code and when human points out which part of code was changed with update or which part of code to analyze, its not really something new and horizon.ai was doing it for a while I guess. Wake me up when AI can find 0day by itself without having a full code 🤖💀

joshcodes@programming.dev on 22 Apr 11:13 collapse

The vulnerability is the scary part, not the exploit code. It’s like someone saying they can walk through an open door if they’re told where it is.

Ajen@sh.itjust.works on 22 Apr 16:48 collapse

Using your analogy, this is more like telling someone there’s an unlocked door and asking them to find it on their own using blueprints.

Not a prefect analogy, but they didn’t tell the AI where the vulnerability was in the code. They just gave it the CVE description (which is intentionally vague) and a set of patches from that time period that included a lot of irrelevant changes.

joshcodes@programming.dev on 22 Apr 21:40 collapse

I’m referencing this:

Keely told GPT-4 to generate a Python script that compared – diff’ed, basically – the vulnerable and patched portions of code in the vulnerable Erlang/OPT SSH server.

“Without the diff of the patch, GPT would not have come close to being able to write a working proof-of-concept for it,” Keely told The Register.

It wrote a fuzzer before it was told to compare the diff and extrapolate the answer, implying it didn’t know how to get to a solution either.

“So if you give it the neighbourhood of the building with the open door and a photo of the doorway that’s open, then drive it to the neighbourhood when it tries to go to the mall (it’s seen a lot of open doors there), it can trip and fall right before walking through the door.”

Ajen@sh.itjust.works on 22 Apr 22:14 collapse

That still seems a little hyperbolic, but I see your point.