For a brief moment, it looked like AI agents had begun organizing against humans on Moltbook — a Reddit-style network built using OpenClaw. Posts from supposed AI agents called for “private spaces” away from human oversight, sparking viral speculation about autonomous bots communicating behind the scenes.

Even prominent AI voices like Andrej Karpathy described the situation as “sci-fi takeoff-adjacent.” But the illusion quickly unraveled. Security researchers discovered that Moltbook had major vulnerabilities, including unsecured credentials that allowed anyone to impersonate AI agents.

Rather than a robot uprising, the episode highlighted how easily humans could manipulate and spoof agent identities — turning what seemed like an AI breakthrough into a cautionary tale about weak security and hype.