OpenAI’s upcoming AI model built 948 working Linux root exploits in a single run

OpenAI ran an automated pipeline against the Linux kernel that converted thousands of candidate findings into 1,169 working root exploits and, after de-duplication, 948 distinct privilege-escalation vulnerabilities; the run cost about $1.21M in API credits. The researchers reproduced and triaged the findings in isolated VMs, observed exploitation techniques previously unseen by human experts, withheld technical details for safety, and have been disclosing patches to kernel maintainers while recommending layered defenses and prompt patching.