When employees and developers adopt new tools independently, IT management often discovers them only after they are deeply embedded in daily workflows. OpenClaw, a widely used AI agent, illustrates this point clearly. On the surface, it provides convenience, managing tasks, sending messages, and automating repetitive workflows. Behind the scenes, it operates with broad authority, holding credentials, executing commands, and connecting across systems without oversight. The vulnerability identified by our Threat Research Team is a window into a larger truth: AI agents are now operational actors rather than simple productivity tools.

Full report : Why AI agents introduce a new security risk to enterprises.