在看过、用过许多个手机/电脑系统级 AI 智能体，也深度使用过「豆包手机」之后，再看这次的 Gemini 智能体，我觉得关于它的讨论不该止于一个「新功能」。
The common pattern across all of these seems to be filesystem and network ACLs enforced by the OS, not a separate kernel or hardware boundary. A determined attacker who already has code execution on your machine could potentially bypass Seatbelt or Landlock restrictions through privilege escalation. But that is not the threat model. The threat is an AI agent that is mostly helpful but occasionally careless or confused, and you want guardrails that catch the common failure modes - reading credentials it should not see, making network calls it should not make, writing to paths outside the project.
，推荐阅读爱思助手下载最新版本获取更多信息
Building the image is done with podman (or docker if you prefer).
The idea is to catch cancer before someone starts to get ill and when it can still be treated.，这一点在heLLoword翻译官方下载中也有详细论述
Жители Санкт-Петербурга устроили «крысогон»17:52。业内人士推荐夫子作为进阶阅读
В Финляндии предупредили об опасном шаге ЕС против России09:28