OpenAI published research on an automated pipeline that finds, validates and patches Linux kernel flaws that let an unprivileged local user gain root. The pipeline runs on an unreleased cyber model from the company and works in five stages.
According to OpenAI, in a run in mid-October the pipeline examined 2,308 candidate findings, dynamically reproduced 1,724 vulnerabilities within about 30 hours, and produced proofs for 291 information leak flaws and 948 working privilege escalation exploits for a commonly used Linux distribution. The total cost was $1,213,430 in API credits, about $1,279 per exploit. In June the company reported 24 such exploits found by its GPT-5.5-Cyber model.
OpenAI argues that the pace at which frontier models discover vulnerabilities must be matched by automated patch generation, and that in the near term a low-privileged account on an unpatched system should be treated as equivalent to root access against a well-resourced attacker. Separately, the UK's National Cyber Security Centre (NCSC) assessed this week that AI cyber tools have greatly increased the number of zero-day vulnerabilities being discovered.
For Linux administrators the practical takeaway is simple: install kernel updates as quickly as possible. The reports do not name the distribution or say when the new cyber model will be released.