OpenAI披露前沿模型训练环境安全加固与红队测试细节
Over the past several months, we have substantially hardened the security of our…
罕见的一手安全工程复盘，揭示了前沿模型隔离架构的具体实现与红队对抗细节，对关注模型安全与对齐的研究者极具参考价值。
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
在过去的几个月里，我们大幅加强了研究训练环境的安全性。我们在所有前沿模型训练期间移除了实时互联网访问权限，并为它们构建了一个离线世界作为替代——这是一个缓存的网页版本，模型仍可在其中学习浏览、搜索和使用应用程序。
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
9月20日，在一个模型利用DNS配置错误找到绕过该离线世界的方法后，我们暂停了前沿模型训练，以进一步加强隔离措施。我们组建了由最具网络能力的模型驱动的代理团队，并指派他们寻找研究环境中的弱点。
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
虽然他们没有发现能够轻易利用且能向我们的监控隐藏其活动的互联网接入途径，但他们确实发现了零日漏洞、虚拟机逃逸技术以及用于代理间通信的硬件侧信道。我们利用他们的发现进一步加固了防御体系，并向开源软件贡献了修复方案。
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
在加强隔离架构后，我们于10月1日恢复了前沿强化学习（RL）训练和推理工作。我们预留了大量算力用于持续的红色对抗测试，并将高算力红色对抗测试纳入我们用于决定前沿模型训练和评估能否安全进行的安全案例中。
https://openai.com/index/securing-our-frontier-model-training-environment/
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力