Proof that AI can now find 0-days even w/o agents or advanced tooling. Also pretty cool: o3 managed to find the vulnerability only 8 out of 100 times. I hadn’t realized it might take that many tries to get a useful result from AI.

sean.heelan.io/2025/05/22/h...

#linux #infosec #llm

The sheer scale of this operation, as well as the investigation, is fascinating www.inversecos.com/2025/02/an-i...

Creating a fuzzer for Chrome’s V8. Down to earth blogpost w/o any illusions apt29a.blogspot.com/2022/01/fuzz...

Fuzzing the Linux kernel: start the campaign, go to sleep, wake up to mysteries you may never solve. Here’s what happened when I took on TIPC network subsystem in Linux: slavamoskvin.com/finding-bugs...
#fuzzing #cybersecurity #pentesting #kernel

Setting up syzkaller and crashing a vulnerable driver: slavamoskvin.com/finding-bugs...

#linux #infosec #fuzzing

This year, I came across many articles, but those really caught my eye:

Google Project Zero's LLM-fuzzing series where they're using LLMs to generate fuzzing test cases
googleprojectzero.blogspot.com/2024/06/proj...
googleprojectzero.blogspot.com/2024/10/from...

I tried to discover the same bug in a linux kernel module with and without KASAN. Here's what's happened: slavamoskvin.com/hunting-bugs...
#linux #fuzzing #cybersecurity