OpenAI security agents uncover nearly a thousand Linux root vulnerabilities

An automated security system running an upcoming OpenAI cyber model built 948 working privilege escalation exploits against the Linux kernel. Operating at roughly 1,279 dollars per exploit, the AI setup proved dramatically cheaper than traditional bug bounty programs while inventing novel hacking techniques. The project supplied candidate security patches alongside detailed vulnerability reports to software maintainers.