OpenAI said its autonomous security agents found 948 distinct local privilege-escalation flaws in the Linux kernel, produced working exploits and drafted fixes for review. The agents examined about 2.8 million lines of code, roughly 7% of the kernel. Exploiting these flaws would first require limited access to a vulnerable machine; human reviewers and software maintainers must still validate and implement any fixes.
OpenAI said the October run cost $1,213,430 in API credits, about $1,279 per flaw. Its earlier July campaign found 166 unique privilege-escalation flaws, while a newer, unreleased model found more than five times as many in the October run.
