How a Tweet Drained $200K From an AI Agent (And Why Threshold Signing Would've Stopped It)

An AI agent just lost $200K because of a single compromised tweet. Not a vulnerability in the AI model. Not a prompt injection attack. An architecture failure. The agent held its own private key. One compromised credential meant full wallet access. Single point of failure, catastrophic result. This