Nope, it's real. OpenClaw has zero filters, zero guardrails, just an LLM with full access to your accounts and APIs with unrestricted access to the web, including reading and processing incoming messages from unknown senders. Attackers can do just about anything with it that they want simply by asking it nicely.