OpenAI's Bot Found 948 Ways to Root Linux: Exploit Writing Is Now a $1,279 Line Item
OpenAI's unreleased cyber model turned 30 hours and $1.2M of API credits into 948 working Linux kernel root exploits - about $1,279 each, and worth $18.9M-$66.3M on the black market. Exploit writing just became a repeatable production line.
Published · By Alfred Belvedere
Today's trend: OpenAI said on November 4, 2026 that a pipeline built around an unreleased cyber model flagged 2,308 candidate bugs in the Linux kernel, reproduced 1,724 of them, and built working root exploits for 948 flaws - meaning an ordinary low-privileged user could take full control of an unpatched machine. The run took about 30 hours in mid-October and cost $1,213,430 in API credits, roughly $1,279 per exploit (SecurityWeek, The Hacker News). At the rates Pwn2Own and TyphoonPWN have paid for kernel exploits, the batch would be worth $18.9M to $66.3M. OpenAI's own write-up says an unpatched system with a low-privileged login should be considered functionally equivalent to root access for a well-resourced threat actor. Within 10 days of the scan, more than 100 of the flaws had been patched or had staged patches from other contributors, with roughly 300 more under public discussion; hundreds of reports and patches have gone to maintainers. On Google's hardened kernelCTF target the same pipeline managed 21 exploits in 112 hours for about $33,000. The signal: exploit-writing has collapsed from a scarce human craft to a priced, repeatable production line - which means every enterprise's defensive budget is now on the wrong side of an elastic supply curve. The moat isn't talent or headcount. It's how fast your team can patch, disclose and verify.