AI 驱动的自动化攻击已进入实战取证阶段,其命令模式可直接做检测线索。
ai_agentllm事件
I've seen AI agents use this pattern quite a lot when running commands. They turn a question into a simple yes/no check and return a predictable result that can be used to decide what to do next.
We're currently investigating a full intrusion where a significant part of the attacker's activity appears to have been automated using AI.
This command is one of the interesting artifacts we've come across. Not proof of AI on its own, but it fits the pattern I've seen. Have a look at the screenshot below.
Maybe I'm biased, but to me, this pattern screams AI.