Microsoft's security researchers have issued a clear warning about the OpenClaw security risk, stating the AI agent runtime is "not appropriate to run on a standard personal or enterprise workstation" because it blends untrusted instructions with executable code while using valid credentials, creating a threat that standard endpoint protection alone cannot block.
