UN Final Report on Governing AI for Humanity

Overview

The document titled 'Governing AI for Humanity' represents the culmination of a year-long intensive study and consultative process led by the United Nations High-Level Advisory Body on Artificial Intelligence (HLAB-AI). Established by Secretary-General António Guterres in October 2023, the Advisory Body was tasked with conducting an analysis and advancing recommendations for the international governance of artificial intelligence. This final report, released in September 2024, addresses the urgent need for a globally coordinated approach to AI, noting that while the technology offers transformative potential for the Sustainable Development Goals (SDGs), it also presents significant risks ranging from disinformation and bias to existential threats and the exacerbation of the global digital divide. The report emphasizes that AI governance is not merely a technical necessity but a moral and political imperative to ensure that the benefits of AI are distributed equitably across all nations, particularly those in the Global South that have historically been excluded from such technological revolutions. The significance of this document lies in its role as a foundational blueprint for the UN's future AI architecture. It was strategically released ahead of the Summit of the Future and the adoption of the Global Digital Compact, serving as the primary advisory input for Member States. The report argues that the current landscape of AI governance is fragmented, with a 'patchwork' of initiatives that often lack the inclusivity required for global legitimacy. By positioning the United Nations at the center of this coordination effort, the report seeks to bridge the gap between existing regional frameworks—such as the EU AI Act or the OECD AI Principles—and the needs of the 118 countries currently not part of any major AI governance initiative. It advocates for a 'distributed' yet 'coordinated' model of governance that respects national sovereignty while upholding universal human rights and international law.

Definitions

The report provides a nuanced conceptualization of Artificial Intelligence and its governance, moving beyond purely technical definitions to encompass the socio-technical nature of the field. It defines AI governance as the set of principles, norms, rules, and decision-making procedures that guide the development, deployment, and use of AI systems. Central to the document’s lexicon is the concept of the 'AI Divide,' which refers to the growing gap between countries that have the computational power, data, and talent to develop AI and those that are merely consumers or are left behind entirely. The report defines 'Global Governance' in this context as an inclusive process that ensures no country is a 'rule-taker' without having had the opportunity to be a 'rule-maker.' Furthermore, the document emphasizes 'Human-Centric AI' as a core definition, signifying AI systems that are designed and operated in a manner that respects human dignity, autonomy, and fundamental rights. It also introduces the concept of 'AI for Humanity' as a guiding star, where the success of the technology is measured by its contribution to the 2030 Agenda for Sustainable Development. The definitions section also clarifies the roles of 'Compute,' 'Data,' and 'Talent' as the three pillars of AI capacity, arguing that governance must address these physical and intellectual resources to be effective. By defining these terms through the lens of international equity, the UN sets a standard that prioritizes collective human well-being over narrow commercial or geopolitical interests.

Governance and Institutional Framework

The institutional framework proposed in 'Governing AI for Humanity' is designed to be lean, agile, and networked, avoiding the creation of a massive new bureaucracy while ensuring robust international oversight. The centerpiece of this framework is the recommendation to establish an International Scientific Panel on AI. Modeled after the Intergovernmental Panel on Climate Change (IPCC), this body would consist of independent experts tasked with providing periodic assessments of AI trends, risks, and socio-economic impacts. This panel aims to create a 'common factual basis' for global policy discussions, ensuring that all Member States have access to high-quality, non-partisan scientific evidence regarding the state of AI technology and its potential trajectories. In addition to the Scientific Panel, the report calls for the creation of a new AI Office within the United Nations Secretariat. This office would serve as a central hub for coordination, reporting directly to the Secretary-General and supporting the various proposed governance functions. It would act as a liaison between Member States, the private sector, and civil society, facilitating the implementation of the Advisory Body’s recommendations. The framework also envisions a 'Global AI Policy Dialogue,' a forum where regulators and policymakers from around the world can meet twice a year to share best practices, align regulatory approaches, and discuss emerging threats. This institutional structure is intended to complement, rather than replace, existing bodies like the ITU or UNESCO, creating a cohesive ecosystem where different UN agencies can contribute their specific expertise under a unified strategic direction.

Key Provisions

The report outlines seven specific recommendations that form the 'Key Provisions' for a global AI governance regime. Beyond the Scientific Panel and the AI Office, a critical provision is the establishment of a Global Fund for AI. This fund would be dedicated to closing the AI divide by financing projects in developing countries that focus on building local compute capacity, developing diverse datasets, and training local talent. The report argues that without financial mechanisms to support the Global South, any talk of 'inclusive governance' remains theoretical. This provision is designed to ensure that AI becomes a tool for convergence rather than divergence in global development. Another major provision is the creation of an AI Standards Exchange. Recognizing that technical standards often act as de facto regulations, the report suggests a mechanism to ensure that developing countries have a seat at the table when these standards are being set by bodies like ISO or IEEE. Additionally, the report proposes a Global AI Data Framework, which would establish international norms for data sharing, privacy, and sovereignty, particularly for data used in training large-scale AI models. These provisions are underpinned by a commitment to the 'UN Charter' and 'International Human Rights Law,' requiring that all AI development and use cases be screened for their impact on fundamental freedoms. The report also highlights the need for 'Capacity Development Networks' to provide technical assistance to governments as they navigate the complexities of AI regulation and adoption.

Scope and Application

The scope of 'Governing AI for Humanity' is intentionally universal, applying to all Member States of the United Nations, as well as non-state actors including private corporations, academic institutions, and civil society organizations. Geographically, the report leaves no region out, but it places a specific emphasis on the 'unrepresented'—the 118 nations that are currently excluded from major AI governance forums. The application of the report’s recommendations is meant to cover the entire AI lifecycle, from the initial research and data collection phases to the deployment, monitoring, and eventual decommissioning of AI systems. This holistic scope ensures that governance is not just a 'downstream' check but an 'upstream' design requirement. In terms of application, the report acknowledges that AI is a general-purpose technology with applications in every sector of human activity. Therefore, while the governance framework is high-level, it is designed to be applicable to specific domains such as healthcare, education, and climate action. The report explicitly states that its recommendations should apply to both 'frontier' AI models (the most powerful and risky systems) and 'everyday' AI applications that impact human rights and economic opportunities. By defining such a broad scope, the UN aims to prevent 'governance arbitrage,' where companies move operations to jurisdictions with weaker oversight, by establishing a global floor for AI safety and ethics that applies everywhere.

Implementation Framework

Implementation of the recommendations in 'Governing AI for Humanity' is envisioned through a multi-stakeholder approach that leverages the existing infrastructure of the United Nations. The primary vehicle for implementation is the Global Digital Compact (GDC), which was negotiated by Member States and adopted during the Summit of the Future in September 2024. The GDC incorporates the Advisory Body’s suggestions into a formal intergovernmental agreement, providing the political mandate necessary for the Secretary-General to move forward with the creation of the AI Office and the Scientific Panel. The implementation framework emphasizes 'interoperability' between different national and regional AI laws, encouraging countries to adopt frameworks that can 'talk' to each other to facilitate cross-border AI innovation. To ensure practical implementation, the report suggests the use of 'AI Sandboxes' and 'Policy Labs' where new governance ideas can be tested in a controlled environment before being scaled globally. The UN Tech Envoy’s office is expected to play a lead role in coordinating these efforts, working closely with regional organizations like the African Union and the ASEAN. The implementation strategy also relies heavily on 'Public-Private Partnerships,' recognizing that the vast majority of AI expertise and infrastructure resides in the private sector. By creating incentives for companies to align with UN principles—such as through the Global AI Fund or participation in the Standards Exchange—the framework seeks to turn high-level norms into daily business practices.

Monitoring and Evaluation

Monitoring and evaluation are built into the proposed UN AI architecture to ensure accountability and responsiveness to the rapid pace of technological change. The International Scientific Panel on AI is the primary mechanism for monitoring, as it will produce an annual 'State of AI' report that tracks technological breakthroughs, emerging risks, and the progress of AI adoption worldwide. This reporting mechanism is intended to provide an early-warning system for 'black swan' events or unforeseen negative externalities of AI systems. Furthermore, the proposed AI Office would be responsible for monitoring the implementation of the Global Digital Compact’s AI-related commitments by Member States and private actors. Evaluation will also occur through the 'Global AI Policy Dialogue,' where nations will periodically review the effectiveness of international governance measures. The report suggests that the governance framework itself should be subject to a 'comprehensive review' every three to five years to ensure it remains fit for purpose in an evolving technological landscape. Metrics for evaluation will include the narrowing of the AI divide (measured by compute and data access in the Global South), the number of AI-related human rights violations reported, and the alignment of AI development with the Sustainable Development Goals. This data-driven approach to monitoring ensures that the UN’s governance efforts are not static but are continuously refined based on empirical evidence.

Relationship to Other Instruments

The 'Governing AI for Humanity' report does not exist in a vacuum; it is designed to be the 'connective tissue' between various existing international instruments. It explicitly builds upon the UNESCO Recommendation on the Ethics of Artificial Intelligence (2021), which provided the first global normative framework for AI. While UNESCO focuses on ethics, the HLAB-AI report focuses on the institutional and political architecture needed to enforce and coordinate those ethics. It also aligns with the OECD AI Principles and the G7 Hiroshima AI Process, seeking to universalize the safety and transparency standards developed by those groups so they can be adopted by the broader UN membership. Furthermore, the document is deeply integrated with the 'Pact for the Future' and the 'Global Digital Compact.' It serves as the technical and advisory foundation for the AI-specific chapters of those documents. The report also references the UN Charter and the Universal Declaration of Human Rights as the ultimate legal authorities to which all AI governance must conform. By referencing these foundational documents, the report asserts that AI governance is not a new field of law but an extension of existing international legal obligations into the digital realm. This relationship ensures that the new AI framework reinforces, rather than undermines, the established international order.

International Alignment

International alignment is a central theme of the report, which argues that the current 'fragmentation' of AI governance is a risk in itself. To achieve alignment, the report proposes a 'Networked Governance' model where the UN acts as a platform for harmonizing different national approaches. This includes aligning definitions of 'high-risk AI' across jurisdictions and creating mutual recognition agreements for AI safety assessments. The report acknowledges the leadership of the European Union with its AI Act and the United States with its Executive Order on AI, but it emphasizes that these domestic policies must be aligned with a global standard to prevent a fractured internet and a bifurcated AI market. The report also highlights the importance of alignment with the Global South. It argues that for AI governance to be truly international, it must reflect the priorities of developing nations, which often focus more on capacity building and economic opportunity than on the long-term existential risks discussed in Western forums. By incorporating these diverse perspectives into the proposed International Scientific Panel and the Global AI Fund, the UN seeks to create a truly global consensus. This alignment strategy is intended to prevent 'digital colonialism' and ensure that the governance of AI is a collective effort of all humanity, as the title of the report suggests.

Implementation Timeline

Adoption and Endorsement

Sources and References