IT之家 · AI 筛选·· 4 hours agoSelectedAI score83
OpenAI Releases Security Report: Scans Linux Kernel to Reproduce 1,724 Vulnerabilities and Generate 948 Privilege Escalation Exploits
OpenAI 扫描 Linux 内核发现 948 个可提权漏洞，2308 个候选问题中有 1724 个成功复现
AI guide
OpenAI has released a cybersecurity model research report detailing the use of an automated agent system to discover vulnerabilities in the Linux kernel, successfully reproducing 1,724 vulnerabilities out of 2,308 candidate issues and generating 948 exploits that achieve local privilege escalation.
Why it matters
The report quantifies the cost and efficiency of next-generation AI models in end-to-end kernel vulnerability discovery and exploitation, providing an empirical reference for defenders assessing the risks of automated offensive and defensive operations.
Source:IT之家 · AI 筛选 · ithome.com