International Legal Foundations of AI Governance and Soft-Law Mechanisms

Abstract

Against the intertwined evolution of globalization and digitalization, the breakthrough development of artificial intelligence (AI) is reshaping international society with unprecedented breadth and depth. The governance challenges it generates transcend the boundaries of any single state, making it urgently necessary to construct a robust and effective international regulatory framework capable of providing systematic responses. At present, the international legal foundations for AI governance remain weak and exhibit a fragmented character. Hard-law mechanisms centered on treaty-making among sovereign states face significant difficulties in forging global consensus due to their inherent procedural rigidity, high negotiation costs, and inability to keep pace with rapid technological iteration. Under these real-world constraints, soft-law mechanisms—comprising guidelines, principles, standards, and codes of conduct advocated and formulated by international organizations, multi-stakeholder forums, professional standard-setting bodies, and industry alliances—are playing an increasingly pivotal, pioneering, and supplementary role in global AI governance by virtue of their flexibility, adaptability, and inclusiveness. This study aims to systematically examine the scope and limits of the effect of existing international-law bases for AI governance and to analyze their structural defects and bottlenecks in application; it then focuses on soft-law mechanisms, offering a detailed account of their core functions, diverse forms, operational logic, and practical influence within today’s governance ecosystem. Finally, the study provides a forward-looking discussion of how soft law and any future international hard-law frameworks might develop dynamic linkages and functional complementarity, with a view to providing a solid theoretical reference and roadmap for building an international cooperative system of AI governance that balances technological development, ethical considerations, security needs, and global public interests. The central thesis advanced here is that, over the medium to long term, in which AI paradigms continue to evolve rapidly, soft-law mechanisms will remain the most active and pragmatic leading normative force in the global governance system, and their successful practice will lay indispensable social-cognitive and normative foundations for any legally binding international regime that may emerge.