Key Info
OpenAI security researchers built an autonomous cyber vulnerability discovery, exploitation, and patching pipeline. In a single run using an upcoming cyber model, it validated and generated patches for 948 local privilege escalation vulnerabilities in the Linux kernel at a total API cost of $1.2 million.
Highlights
- The agents discovered and used at least 10 Linux exploitation techniques that OpenAI and consulted experts believe are novel.
- This suggests frontier LLMs can discover and patch new exploitation approaches, not just apply known techniques.
- OpenAI says the goal is to help defenders find and fix vulnerabilities before attackers exploit them.
- The team is working with Linux maintainers and Patch the Planet partners on reports and patches.