Last month, Elon Musk’s DOGE team outlined a plan to transform the United States federal government through an “AI-first strategy.” It is a typical story that engineers walk into government departments with swashbuckling confidence, believing that technical expertise can break through existing bottlenecks and solve all the government’s problems. Given this viewpoint, DOGE’s AI developers will surely be more worried about avoiding false negatives (overlooking an instance of fraud) than false positives (incorrectly labeling a contract as wasteful or fraudulent), particularly in cases where they may have a conflict of interest.