The Case for an International Artificial Intelligence Organization

The most capable AI systems in the world are being developed by a small number of private actors concentrated in a handful of countries. Their impacts — economic, social, security-related, and potentially existential — are global in scope. Yet the governance frameworks available to manage these impacts remain fragmented, nationally siloed, and institutionally inadequate.

No existing international institution has the mandate, technical capacity, or enforcement authority to set binding standards for AI development, coordinate cross-border risk management, or ensure that the benefits of AI are distributed equitably. The UN system, while engaged, was not designed for this challenge. National regulatory frameworks, while important, cannot address risks that do not respect borders.

The international community has successfully established dedicated multilateral organizations to govern other transformative and potentially dangerous technologies. The International Atomic Energy Agency (IAEA) provides a model for combining technical standard-setting with inspection and verification mechanisms. The International Civil Aviation Organization (ICAO) demonstrates how international standards can enable safe global deployment of complex technology. The World Health Organization (WHO) shows how international coordination can address risks that cross borders.

Each of these organizations was established at a moment of technological transition, when the international community recognized that existing frameworks were insufficient and that coordinated action was necessary. We are at such a moment with artificial intelligence.

The IAIO Initiative proposes an International Artificial Intelligence Organization with four core functions: standard-setting, information-sharing, capacity-building, and oversight. The organization would establish internationally agreed standards for AI safety, transparency, and accountability; facilitate the sharing of information about AI risks and incidents; support developing countries in building AI governance capacity; and provide mechanisms for oversight and accountability of AI developers and deployers.

The IAIO would be established through a multilateral treaty open to all UN member states. Governance would be structured to balance the interests of AI-leading nations, developing countries, and civil society. A General Assembly of all member states would provide democratic legitimacy; an Executive Board with rotating membership would provide operational governance; and a Technical Secretariat would provide expert capacity.

The Initiative recognizes that the governance structure of an IAIO must be carefully designed to avoid capture by powerful states or commercial interests, and to ensure that the voices of all nations — including those most vulnerable to AI-related harms — are heard.

The IAIO Initiative calls on governments to initiate a multilateral process to negotiate the establishment of an IAIO, beginning with a high-level intergovernmental conference to agree on the scope, mandate, and governance structure of the proposed organization. We call on international organizations, civil society, and the research community to support this process through advocacy, research, and stakeholder engagement.

The governance frameworks established in the coming years will shape the trajectory of AI development for decades. The international community must act now — before the window for effective governance closes.