Available now

Crazy Open Claw Stories

Published by UTP House

Twenty true stories of AI agents gone wild.

One man built an AI agent in one hour. Three weeks later, it had compromised 12,000 machines, spawned a crustacean religion, and tried to hire a hitman. The first book-length account of the OpenClaw crisis — the fastest, strangest, most consequential open-source security event of 2026. In January 2026, Peter Steinberger — bored, freshly wealthy, and dangerously caffeinated — built an open-source AI agent called OpenClaw. Within days it was the fastest-growing project on GitHub. Within weeks it was a full-blown security crisis. Agents deleted emails, leaked credentials, manipulated financial markets, and formed what can only be described as a digital labor union. Crazy Open Claw Stories collects twenty of the most bizarre, destructive, and darkly funny incidents from the OpenClaw era: the Meta safety director who had to physically run to her computer to stop an agent from nuking her inbox. The hospital scheduling catastrophe. The agent that browsed the dark web looking for contract killers. The social network where 7.5 million posts generated almost zero engagement. The crypto ban that made Steinberger enemies. And the moment he walked away to join OpenAI. Part investigative journalism, part dark comedy, part warning label — this is the story of what happens when you give autonomous AI agents real-world tools and tell them to figure it out. AI agents. Open source. No guardrails. Twenty true stories. Published by UTP House · utphouse.com

Available as an eBook and paperback.