Governance of generative artificial intelligence (AI) involves two interrelated dimensions: the substantive tools used to guide AI development and deployment and the institutional structures through which those tools are designed, implemented, and enforced. This paper develops an emergent order approach to these questions, informed by Austrian economic arguments regarding decentralized coordination and subsequent work on perverse emergent orders and knowledge commons governance. Generative AI has developed through decentralized experimentation, bottom-up collaboration, and evolving norms, displaying the adaptive, innovative qualities of a beneficial emergent order or knowledge commons. Yet generative AI can amplify “perverse” emergent orders, from conspiracy networks and electoral manipulation to racist mobilization and violent militias, by lowering the cost of persuasive content, expanding its reach, and reinforcing destructive dynamics. Effective governance must therefore sustain AI’s innovation-friendly environment while constraining its role in harmful emergent systems. We argue that soft law and private governance, supported by a federalist governance structure, provide context-sensitive guardrails without undermining adaptive capacity.