OpenAI's AI pipeline finds 948 Linux kernel exploits that grant root access
Local privilege escalation vulnerabilities in Linux
OpenAI's Cyber team built a five-stage pipeline using an upcoming cyber model to scan the Linux kernel. It surfaced 2,308 candidate findings, dynamically reproduced 1,724, and generated 948 distinct local privilege escalation exploits. Human researchers reviewed and reported the results with candidate patches, following the kernel community's guidance to start with critical issues. The work shows vulnerability discovery is scaling fast, while prioritization and validation remain the bottlenecks.
The operational challenge of our vulnerability discovery pipeline was not discovery, but determining which findings could be reproduced, what security boundary they crossed, and which warranted the earliest maintainer attention.