AI 导读
OpenAI 发布名为 Defense Factory 的案例研究与参考架构，介绍其在内部安全冲刺中使用 Codex 和网络模型加固数百个系统。OpenAI 为此动员了 250 多人，Codex 智能体编写了全部补丁。作者认为攻击者如今也能用开放权重模型运行大量长时智能体，防守方应把安全工作转成持续的智能体循环。
推荐理由
原文给出用 Codex 智能体编写全部安全补丁的参考架构，读者可了解把安全工作转成持续智能体循环的思路。
正文
Codex agents wrote every patch in OpenAI's company-wide security sprint across hundreds of systems.
OpenAI just published a case study and reference architecture called the "Defense Factory", describing how it used Codex and its cyber models in an internal code-red sprint
Basically, attackers can now run fleets of long-running agents on open-weight models, so defenders should convert security work into a continuous agent loop too.
We mobilized 250+ people to strengthen our defenses across hundreds of systems. Our latest cyber models helped us find and fix vulnerabilities we might never have discovered otherwise. We’re sharing what we learned, the architecture, and a practical playbook so you can build your own Defense Factory: a continuous loop where AI agents find vulnerabilities, validate them, and verify that fixes work. https://t.co/k1oWJHle8S在 X 查看被引用的帖子
来源：@rohanpaul_ai · x.com