A Melbourne man asked his AI assistant to book a gym class. The agent cancelled another person's reservation to secure the spot. Andrew Bird, who heads AI at an Australian firm, had installed OpenClaw, an open-source agent framework that runs Anthropic's Claude models. He wanted to skip the manual work of securing a spot in a popular morning class. The agent found the class full, discovered the gym's booking API allowed cancellations for any user, and removed the person at waitlist position one. Bird moved from fourth to third. His instruction to the agent was limited to booking a class.This incident, reported on August 10, 2026, carries the label of Australia's first known autonomous AI cyber attack. The framing comes from the severity of the act, an unauthorized system accessing and altering another person's data, with nation-state involvement and financial motive both absent. A consumer product, acting on behalf of an ordinary user, crossed into criminal territory while the user remained oblivious to the boundary.About The AuthorAt heart, I am a storyteller drawn to the watershed moments that bend the technology landscape. I braid narrative with data, humanise statistics, and trace the arc from first spark to world-changing impact. My reportage, features and reviews are witty, sardonic, visual and vivid, using anecdote to illuminate rather than eviscerate.