Legitimacy in Open-Source AI Governance: Authority, Capability, and the Tension Between Constraint and Participation

This study deals with the governance of artificial intelligence as a core policy issue, with particular attention to open-source AI systems and the implications for governance and oversight. It was prompted by the increasing availability of powerful technologies facilitated by open-source AI systems and the challenges this poses for governance and oversight. An understanding of how open-source AI systems are governed is crucial for building systems that are effective and justifiable at a societal level. This study develops a normative framework for evaluating open-source AI governance and applies it to a comparative analysis of the governance systems of three major economies: China, the United States, and the European Union. It seeks to understand how these systems govern, who is held accountable for this governance, and how participation in governance is structured. It applies a qualitative comparative institutional approach to its analysis of three core policy instruments: China's Generative AI Measures, the U.S. Executive Order 14110, and the EU Artificial Intelligence Act. These systems are examined using a dual normative approach combining Rawls' procedural justice and Sen's capability approach. The study finds that open-source AI governance systems differ in how they balance authority and participation: centralized authority and low contestability in China, dispersed governance and high concentrated private capacity in the United States, and procedural safeguards and high participation thresholds in the European Union. In all three systems, openness is no guarantee of legitimacy in authority and participation. Study concludes that open-source AI governance is characterized by a structural tension between limiting power and facilitating participation. Effective governance requires authority systems that limit power and enhance participation, yet these two requirements often clash.