发生了什么: Safety overview: GPT-6 Astra

OpenAI's GPT-6 Astra model reached Critical cybersecurity capability level under the Preparedness Framework.

为何重要: Safety overview: GPT-6 Astra

Security teams and AI safety researchers should track this milestone as it signals advanced model capabilities in security domains.

The source provides minimal detail on what Critical level means, specific capabilities, or mitigation measures implemented.

新闻

OpenAI · 发布于 2026年9月3日 · 阅读约3分钟

发生了什么: OpenAI's GPT-6 Astra model reached Critical cybersecurity capability level under the Preparedness Framework.
为何重要: Security teams and AI safety researchers should track this milestone as it signals advanced model capabilities in security domains.
注意: The source provides minimal detail on what Critical level means, specific capabilities, or mitigation measures implemented.

每个模式都讲清楚技术如何运作、何时值得投入，以及在哪里会失效。

每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。

每周一封邮件,一键退订。您的地址仅用于发送简报。