Artificial Intelligence as a Geopolitical Legal Actor: Rethinking International Law for the LLM Era

Abstract

The rapid diffusion of large language models (LLMs) has transformed artificial intelligence from a technical tool into a strategic infrastructure capable of shaping state capacity, private power, information environments, and the practical operation of legal norms. This article examines whether, and in what analytical sense, artificial intelligence can be described as a geopolitical legal actor in the LLM era. It does not argue that AI systems possess legal personality. Rather, it proposes that frontier AI systems function as actor-like socio-technical infrastructures because they mediate legal obligations, redistribute bargaining power, influence public reason, and alter the conditions under which states exercise sovereignty, jurisdiction, responsibility, and human-rights duties. Using normative legal research, document analysis, and public empirical indicators from the Stanford AI Index, the European Union, the Council of Europe, the United Nations, UNESCO, OECD, NIST, the G7 Hiroshima Process, and other governance instruments, the article identifies a central gap: existing international law recognizes states and corporations as legal subjects and duty-bearers, but it lacks a coherent framework for AI systems that operate across borders, rely on concentrated compute and data supply chains, and are deployed through globally dominant platforms. The article advances a novelty claim by connecting geopolitical AI concentration, international legal fragmentation, and LLM-mediated governance into one framework. It argues for a layered international legal architecture combining human-rights impact assessment, compute and model transparency, cross-border accountability, remedies for affected persons, and inclusive capacity-building for developing countries.