While I'm highly interested in exploring the limits of AI, I foresee enormous problems when virtually limitless AI-powered systems are running in my company's IT infrastructure. The team at SANS has written a practical tool to at least detect Clawdbot/Moltbot/OpenClaw.