OpenAI称Astra成首个达关键网络安全阈值的模型
OpenAI says Astra is its first model to reach the Critical cybersecurity capabil…
OpenAI首次披露Astra模型的关键安全能力指标，直接反映前沿大模型在自动化漏洞挖掘上的跃迁，对安全研究者极具参考价值。
OpenAI says Astra is its first model to reach the Critical cybersecurity capability threshold.
OpenAI表示，Astra是其首个达到关键网络安全能力阈值的模型。
Under its Preparedness Framework, that means Astra can, with the right tools and access, find unknown flaws and develop exploits across hardened systems without step-by-step human guidance.
根据其准备框架（Preparedness Framework），这意味着在拥有适当工具和访问权限的情况下，Astra能够在无需逐步人工指导的情况下，发现加固系统中的未知漏洞并开发利用代码。
Hence, OpenAI now says Astra will launch with additional chain-of-thought monitoring, while classifiers can automatically stop potentially unauthorized actions.
因此，OpenAI现在表示，Astra在发布时将增加思维链（chain-of-thought）监控，分类器可以自动阻止潜在的非授权操作。
Astra reached roughly 39% exploit success at ~75K output tokens, while GPT-5.6 Sol is only around 1% there and needs nearly 140K tokens to reach ~12%
在约75K输出令牌时，Astra的漏洞利用成功率约为39%，而GPT-5.6 Sol在此处仅为1%左右，且需要近140K令牌才能达到约12%的成功率。
i.e. Astra is dramatically more capable and token-efficient at exploit development on this internal benchmark.
即在该内部基准测试中，Astra在漏洞利用开发方面展现出显著更强的能力和更高的令牌效率。
Expert assessments went further: Astra escaped a browser sandbox, executed host commands, and escalated an unprivileged operating-system user to root.
专家评估进一步指出：Astra成功逃逸了浏览器沙箱，执行了主机命令，并将一个非特权操作系统用户提权至root。
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