These numbers from OpenAI's Linux kernel security research are wild. 💀
2,308 candidate findings.
1,724 validated.
948 working local privilege-escalation exploits.
And 291 working information-leak proofs of concept.
All from an automated security research pipeline.…