The technologies of artificial intelligence are rapidly advancing, and many experts in both technical and policy domains fear the emergence of a machine with powers vastly dwarfing humanity, and capable of extinguishing humanity. These fears, as well as a variety of shorter-term systemic risks, have caught the attention of governments, policymakers and the public imagination, who are now scrambling to understand and regulate this new frontier. Human survival, as well as the protection of many lesser human ends, may depend on the design and implementation of restraints on AI. But can this be done? Owing to its world-historic novelty, the prospect of controlling AI is forestalled by a basic confusion and widespread political disagreement about the fundamentals of the problem, even as the pace of change accelerates. This paper explores how the seemingly novel ASI control problem in many ways extends and parallels the republican-constitutional project of restraining centralized state power to protect popular liberty and security, and of designing machines, infrastructures, and artifacts for public safety. By reframing this entire debate through the lens of republicanism and contextualizing this new frontier as an acceleration of historical problematiques, we aim to ground this conversation as well as suggest a path forward based on the framework of restraint.