Thursday, November 5, 2026 · ChatGPT & OpenAI
OpenAI · Safety · happened Nov 4, 2026
OpenAI says an agent pipeline on an unreleased cyber model produced 948 working Linux root exploits for about $1.2 million
OpenAI published results from a mid-October run in which agents on an upcoming cyber-focused model searched the Linux kernel for bugs, reproduced 1,724 of 2,308 candidate findings, wrote 948 working local privilege-escalation exploits for a common distribution, and drafted patches. The run cost about $1.21 million in API credits, roughly $1,279 per exploit, and security researchers review findings before they go to kernel maintainers. OpenAI has not named the distribution or said how many bugs have been reported or fixed so far, and the results have not been independently checked.
Why it matters
It shows how quickly AI can now find and weaponize kernel bugs, which raises the pressure on patching to keep up.