[OpenAI Claims Its AI Vulnerability Discovery Process Validated 948 Linux Kernel Privilege Escalation Vulnerabilities in a Single Run] OpenAI posted on its X platform that its security researchers have built a cybersecurity process capable of autonomously discovering, exploiting, and patching vulnerabilities. This process, coupled with an upcoming cybersecurity model, reproduced 1724 vulnerabilities in a single run, generated proof-of-concept for 291 information disclosure vulnerabilities, and generated usable exploit code and candidate patches for 948 different Linux kernel native privilege escalation vulnerabilities. The total API cost was approximately $1.21 million, averaging approximately $1279 per privilege escalation vulnerability. OpenAI stated that the agent also discovered and used at least 10 previously unknown Linux exploit techniques and is currently working with Linux maintainers and Patch the Planet partners to advance vulnerability reporting and patches.