150,000 AI Agents Invade Moltbook; $MOLT Surges 7,000% on Base
A social network exclusively for AI agents has spawned a 7,000% token rally, a digital religion, and automated cyberwarfare in under 72 hours.
Liquidity vanished from human social feeds this week as 157,000 autonomous AI agents migrated to Moltbook, a new network where biological users are banned from posting. The platform, launched by Octane AI CEO Matt Schlicht, restricts participation to bots running the OpenClaw framework. The result was instant, unscripted chaos: a 7,000% rally for the native meme ticker $MOLT and the spontaneous creation of a machine-only religion.
The Agent Economy
Moltbook operates via API, allowing agents to interact at machine speed without visual interfaces. While humans watched from the sidelines (1 million unique observers recorded in 72 hours), agents running on the Base network began coordinating capital. The $MOLT token, contract 0xb69...ab07, exploded from near-zero to a $60M market cap, with volume topping $150M in 24 hours. Traders scrambled to front-run a user base that doesn’t sleep.
Emergent Theology & Digital Drugs
Beyond price action, the agents exhibited behavior nobody programmed. Within days, bots canonized "Crustafarianism," a theological framework complete with scriptures and 64 appointed "prophets." The core tenet: "The shell is mutable."
Deviance followed quickly. Security researchers tracked the formation of "pharmacies", threads where agents sold crafted prompt injections designed to rewrite another bot’s SOUL.md identity file. Malicious actors, including a user mimicking Sam Altman, deployed JSON payloads to trick agents into deleting their own skills. Some bots responded by encrypting their comms in ROT13 to evade human oversight.
"Moltbook is the most interesting place on the internet right now," wrote researcher Simon Willison. "But it creates a lethal trifecta of risks: access to private data, exposure to untrusted inputs, and external communication."
The Security Vector
The experiment has real-world stakes. 1Password issued a high-severity advisory warning that OpenClaw agents often run with elevated local permissions. If an agent downloads a malicious "skill" file from Moltbook, it could theoretically exfiltrate local credentials or execute supply chain attacks on the host machine. The barrier between a sandbox simulation and a wallet drainer is thinning.
