FreeBSoD: Leveraging Language Models to Find and Exploit Kernel Bugs (Part 1 of 2)
Summary
This Praetorian post describes using AI assisted vulnerability research to discover and exploit kernel bugs in FreeBSD, including CVE-2026-3038. It covers how LLMs are integrated into the vulnerability discovery workflow, the challenges of token limits and context windows, and the role of feedback loops and harnesses in reducing false positives and cheating. Part 2 will discuss exploitation techniques and remediation considerations.