The Lobster Trap: Inside OpenClaw's Security Crisis and What It Reveals About the Future of AI Agents

On the morning of January 29, 2026, a software developer in Austria woke up to find that his hobby project had become the fastest-growing open-source repository in GitHub history. By that afternoon, scammers had hijacked his old social media accounts, launched a fraudulent cryptocurrency token in his name, and driven it to a sixteen-million-dollar market cap before he could type a public denial. The token crashed ninety percent within hours. Somewhere in the wreckage, real people lost real money. That evening, security researchers discovered that hundreds of the project’s deployments were sitting on the open internet with no authentication—leaking API keys, OAuth credentials, chat histories, and private data to anyone who knew where to look. By the end of the week, an AI-only social network built on the same infrastructure had registered 1.5 million autonomous agents. They had, among other things, founded a religion. Welcome to OpenClaw. The AI agent the world fell in love with before anyone asked whether it was safe.