Source-linked AI summary
Regulating Artificial Intelligence: Proposal for a Global Solution
Olivia J. Erdélyi, Judy Goldsmith
TL;DR
AI’s global reach and diverse risks expose limits in fragmented national regulation and create challenges for designing effective governance. The paper analyzes transnational lawmaking and proposes a coordinated international framework, potentially organized around a new or repurposed intergovernmental organization. It concludes that legitimacy, consistency, interdisciplinary expertise, and stakeholder representation are necessary for workable AI regulation, while transnational norms remain vulnerable to fragmented implementation.
Problem
AI’s global reach, substantial risks, and fragmented domestic approaches make effective regulation difficult, while misconceptions about regulation impede contemporary policy efforts.
Method
The paper analyzes transnational legal ordering, institutional architecture, and communication challenges to develop a coordinated international AI governance proposal.
Results
The paper proposes an international AI governance framework centered on a new or repurposed intergovernmental organization to coordinate national policymaking and develop consistent standards.
Takeaways & Limitations
Effective AI governance should align domestic and international arrangements, involve affected stakeholders, draw on interdisciplinary expertise, and treat regulation as a comprehensive process rather than rulemaking alone.
Takeaways & Limitations
Transnational norms can produce fragmented implementation when influential domestic actors are excluded or agreements contain ambiguity and contradictions.
Abstract
from arXiv · showhide
With increasing ubiquity of artificial intelligence (AI) in modern societies, individual countries and the international community are working hard to create an innovation-friendly, yet safe, regulatory environment. Adequate regulation is key to maximize the benefits and minimize the risks stemming from AI technologies. Developing regulatory frameworks is, however, challenging due to AI's global reach and the existence of widespread misconceptions about the notion of regulation. We argue that AI-related challenges cannot be tackled effectively without sincere international coordination supported by robust, consistent domestic and international governance arrangements. Against this backdrop, we propose the establishment of an international AI governance framework organized around a new AI regulatory agency that -- drawing on interdisciplinary expertise -- could help creating uniform standards for the regulation of AI technologies and inform the development of AI policies around the world. We also believe that a fundamental change of mindset on what constitutes regulation is necessary to remove existing barriers that hamper contemporary efforts to develop AI regulatory regimes, and put forward some recommendations on how to achieve this, and what opportunities doing so would present.
1 Introduction
AI is pervasive and transformative, but its benefits and risks create an urgent need for effective regulation. The paper argues that AI governance must address international coordination, institutional design, and communication barriers.
- AI applications span transportation, medicine, finance, justice, surveillance, and weapons, affecting virtually all aspects of human society.
- AI creates risks through biased outputs, privacy threats, human de-skilling, ethically questionable uses, and potentially dangerous autonomous systems.
- Effective AI regulation should harness benefits and address risks proactively while supporting trust, innovation, safety, and socially optimal outcomes.
- Because AI-related externalities cross borders, fragmented national approaches are inadequate and require coordinated transnational regulation.
- Regulation comprises detection, rulemaking, supervision, enforcement, assessment, and adaptation rather than rules alone.
- The paper proposes coordinated national, regional, and international governance frameworks, potentially centered on an international AI organization, while emphasizing interdisciplinary expertise and stakeholder collaboration.
2 Dynamics of Transnational Lawmaking
Transnational AI governance depends on choosing suitable legal and institutional forms and managing recursive interactions between international and domestic systems. Legitimacy, representation, clarity, and intermediaries shape whether transnational norms produce durable behavioral change.
- Transnational legal ordering includes norms and institutions spanning national boundaries, encompassing treaties, standards, guidelines, organizations, and networks.
- Hard and soft legalization involve context-dependent tradeoffs among precision, obligation, delegation, sovereignty costs, flexibility, and implementation capacity.
- Actors may favor hard law when consensus, monitoring, enforcement, collective oversight, or routine administration make formal institutions valuable.
- Transnational norms can produce fragmented implementation when influential domestic actors are excluded, agreements remain vague, or unresolved contradictions permit conflicting national interpretations.
- State change occurs through recursive vertical interactions between transnational and domestic venues and horizontal interactions among transnational legal organizations.
- Norms are more likely to influence behavior when legitimate, clear, coherent, and well understood, while intermediaries help coordinate communication between national and transnational levels.
- AI governance frameworks should be consistent across levels and represent affected actors so resulting rules gain acceptance.
3 Proposal for a New International Artificial Intelligence Organization
The paper proposes international AI cooperation through an intergovernmental organization, initially using flexible soft-law arrangements while drawing on existing initiatives and expertise. It argues that legitimate governance architecture should precede uniform transnational AI rules.
- Proposal and institutional alternatives: The proposed IAIO would provide a focal point for international AI policy debates and could gain regulatory responsibilities as support grows.The authors also identify OECD.AI, UNESCO’s AI group, GGAR, and GPAI as possible alternatives to creating a new organization.
- Institutional design tradeoffs: Because AI affects national security, rights, ethics, and rapidly changing legal questions, the paper favors flexibility before binding international commitments.Flexible cooperation is intended to help states develop familiarity, resolve differences, and establish common ground.
- Institutional design tradeoffs: The paper evaluates institutional choices across delegation, information control, contracting costs, administration, and uncertainty management.It argues that high sovereignty costs and unresolved questions about purpose, membership, and regulatory scope make a gradual approach politically realistic.
- Initial institutional form: The IAIO should initially use soft-law recommendations, guidelines, and standards to encourage internationally consistent national AI policies.The proposed initial form is an informal intergovernmental organization with limited institutional formality, potentially hosted by a neutral country.
- Existing initiatives: Existing initiatives have expanded since the proposal was first conceived, creating possible foundations for an international AI policy body.The discussion notes OECD.AI, GGAR, and other organizations as developments relevant to the proposed architecture.
- Governance legitimacy: Legitimate governance arrangements should precede rulemaking because the quality and legitimacy of the framework determine whether transnational AI norms are efficient and widely accepted.The paper therefore treats governance design as a prerequisite for effective international AI rules.
4 Communication: An Insidious Regulatory Challenge
The paper identifies communication and coordination barriers among governments, academia, business, and civil society as obstacles to effective AI regulation. It recommends clearer objectives, better education about modern regulation, and more coordinated stakeholder engagement.
- Modern regulation: Modern AI regulation is a co-produced process in which governments and autonomous social actors jointly construct knowledge and exercise regulatory power.The paper contrasts this with the outdated view that regulation is solely state-imposed binding rules.
- Coordination failures: Poor coordination and unclear objectives have prevented much stakeholder energy from becoming concrete, implementable action.The paper reports declining enthusiasm in some venues as productivity remains low and AI hype wanes.
- Recommendations: Self-regulatory features may help create regimes with well-defined objectives that coordinate stakeholders more efficiently.The paper presents this as an urgent option alongside broader regulatory approaches.
- Recommendations: Governments and international bodies should set clear, realistic agendas and educate participating actors about regulation and current regulatory practices.The paper calls for a mindset shift from viewing regulation as state-imposed restriction toward recognizing interdependence among regulatory actors.
- Communication barriers: Divergent interests, expertise, and working methods create communication problems that hinder the transfer of knowledge needed to address AI challenges.The paper specifically describes weak understanding between disciplines and gaps between research, government, and business practice.
- Consequences: These communication and coordination problems reduce engagement, trust in regulators, and the acceptance and legitimacy of AI rules.The authors connect improved collaboration and more responsible judgments about expertise with technically feasible policy solutions.
5 Conclusion
The paper highlights key regulatory problems and proposes an international framework, centered on a new or repurposed intergovernmental organization, to coordinate national AI policymaking. Its objective is to prevent fragmented national policies that could generate international tensions.
- The paper identifies key regulatory considerations and problems to support domestic and international AI policymakers.
- It proposes a consistent international regulatory framework centered on a new or repurposed intergovernmental organization to streamline and coordinate national policymaking.
- The framework aims to avoid nationally fragmented AI policies, which may lead to international tensions.