Governing AI in a Multipolar World: A Moral Parliament Grounded in Palauan Polycentric Governance

This paper proposes a novel conceptual framework for AI governance in a multipolar world, drawing on Palau’s polycentric governance tradition to establish a global moral parliament. It responds to calls from Dan Hendrycks’ AI Safety, Ethics, & Society (2024) on the question of “Can the production of AI technology be controlled?” It includes a larger discussion from scholars in the emerging AI governance field (as distinct field of research). Unlike technocratic or top-down models, it operationalizes epistemic justice by integrating diverse knowledge systems—Indigenous, scientific, and pluralistic—through institutional mechanisms into an institution design of a two-chamber AI governance global parliament. The architecture consists of a complex design based on today’s reality of multiple centers of powers and diverse stakeholders, including the Global Majority. It attempts to balance these powers and interests by drawing from Palau’s indigenous governance system that has successfully navigated multipolarity, using a polycentric system, for centuries. It emphasizes teeth over norms, where persuasion escalates to temporary exclusion, ensuring accountability without coercion. It acknowledges limitations and areas for future research and development. Key contributions: Grounds global AI safety in indigenous Pacific Islander political philosophy Offers practical design for multipolar compute governance and epistemic inclusion of the Global Majority Bridges Indigenous normative frameworks with AI policy realism Ideal for: AI governance researchers, policymakers, indigenous scholarship researchers, Global Majority researchers, decolonization scholars, and Pacific/Oceania scholars and researchers. February 2026 preprint.