When will an Artificial Intelligence run for elected office? This may seem like a strange provocation– just an invitation to futurism and speculation. Yet Artificial Intelligence (AI) systems are rolling out over many domains of economic, cultural and political life. They handle public policy and legal issues that humans find complex, mundane, time-consuming, or easily corrupted. They are purposefully designed to make the best possible decisions from vast quantities of information—we might hope a Prime Minister or President would adjudicate this way. But we need to move past abstract ethical problems, and be aware that we are choosing to apply AI systems to tangible social problems and politically sensitive issues. I argue that AIs are political by design, through the political values of their designers and the political nature of training data. AIs will validate some political ideas, and be pressed into service for political ideologies. In this talk, I show how AI is changing our experience of politics and rewriting democracy’s “terms of service”.