The pipeline finds security-relevant code, scans for vulnerabilities, reproduces them in isolated environments, tests whether they can be exploited, and generates candidate patches.

Human researchers review the results before reporting vulnerabilities and propose fixes to

media

原文链接:https://x.com/OpenAI/status/2118119454681006460

中文翻译(AI 生成,仅供参考)

该流水线会寻找与安全相关的代码,扫描漏洞,在隔离环境中复现这些漏洞,测试其是否可被利用,并生成候选补丁。

人工研究人员在报告漏洞之前会审查结果,并提出修复方案。