The First Global Dialogue on AI Governance: Who Gets to Govern AI?
Author:Arindam Bhattacharya
Published: September 15, 2026
|Journal: Global Perspectives on Politics and Diplomacy
|ISSN: 3050-4589
|Abstract
The inaugural session of the United Nations Global Dialogue on AI Governance, convened in Geneva on 6–7 July 2026, was the first annual session of a universal UN platform for discussion of international cooperation on AI governance. This paper uses the Geneva Dialogue not as an event to be reported but as a diagnostic case through which to examine a deeper political question: who actually possesses the power to govern artificial intelligence? Drawing on official United Nations documents, the preliminary report of the Independent International Scientific Panel on Artificial Intelligence, and contemporary institutional and statistical evidence available before the publication deadline, the analysis distinguishes seven analytically separate dimensions of governing power—participation, representation, influence, agenda-setting, decision-making, implementation capacity and technological power—and argues that international AI-governance practice and commentary can conflate procedural participation with substantive governing agency. The paper develops an original Advocacy Unified Network (AUN) framework of “AI Governance Agency,” understood as a bundle of epistemic, normative, standard-setting, infrastructural, regulatory, implementation and accountability capabilities. It then maps the global distribution of these capabilities across compute, semiconductors, data, models, cloud, talent, standards, financing and regulatory capacity, and tests the proposition that universal participation does not automatically produce universal governing agency. The findings indicate that the Dialogue’s universalism is procedurally substantial but does not by itself redistribute the capabilities that make a seat meaningful. The paper concludes with concrete recommendations at global, regional, national, technical and civil-society levels, and proposes an institutional test for success that the Dialogue must meet if it is to become more than another layer in an already fragmented governance landscape.