HomePostsDigital RightsA UN Institution for AI Governance: why and how it could be...

Related Posts

A UN Institution for AI Governance: why and how it could be a good idea

Reading Time: 9 minutes
Print Friendly, PDF & Email

In the last six months, the United Nations has taken the driver’s seat in the debate on the global governance of Artificial Intelligence. Fuelled by the increasing interest about the potential benefits and dangers of AI, the UN stepped in with the aim of harmonising standards and providing a robust AI governance architecture to protect human rights and advance the realisation of the Sustainable Development Goals (SDGs). While the UN is certainly in a privileged position to undertake this role, there is a need for a dedicated institution to be effective.

In December 2023, the AI Advisory Body of the United Nations released an interim report on artificial intelligence, “Governing AI for Humanity”, highlighting the key principles and functions for a UN governance of AI. The UN approach is underpinned by a strong commitment to foster AI for enabling the realisation of the SDGs, sharing the benefits of AI globally and ensuring that technological development happens without undermining human rights. 

On 21st March 2024, the UN General Assembly adopted a Resolution on AI and Human Rights, titled ‘Seizing the opportunities of safe, secure and trustworthy artificial intelligence systems for sustainable development’,  that echoed the Advisory Body’s concerns and recommendations on the importance of leveraging AI to support developing countries, achieve the realisation of the SDGs and protect human rights. Moreover, it explicitly mandates the UN and all its agencies to work to this end.

This adds to the UN Global Digital Compact, whose zero draft was released on 1st April 2024, and confirms the UN commitment to be a protagonist in the regulation and governance of new and emerging technologies.

1. The idea of a UN AI institution 

Despite the criticisms often moved to the UN for its inability to enforce international law, it is probably still the best option for a global governance of AI, given its global authority, experience and commitment to fostering international cooperation. Investing the UN in the global governance of AI could include providing a platform for international dialogue, delegating the drafting of common guidelines and frameworks, and overseeing their enforcement globally. However, all this would require the establishment of an ad hoc institution. Such a body, situated within the broader UN framework, would be essential for effective global AI governance. 

Establishing a new institution is always a challenge and needs a complex series of compromises and commitment of resources and political will. Yet, it appears a necessary evil if the UN wishes to implement all the functions listed in the UN AI Advisory Body report. Following the General Assembly Resolution, Marc Roterberg has suggested the establishment of a dedicated UN Special Procedure Mandate on AI and Human Rights. While certainly helpful and easier to implement, this won’t be able to fulfil all the functions the UN commits to undertake. 

The UN AI Advisory Body report lists seven functions that a global governance of AI would entail. They include: horizon scanning and building consensus; ensuring interoperability and alignment with norms; mediating standards, safety and risk management frameworks; facilitating the development and use of AI; promoting international collaboration on data, computing and talent to solve SDGs; reporting and reviewing; and elaborating norms, ensuring compliance and accountability. The Report recognises that the ‘institutional hardness’ increases with the functions and for the last ones, there is a need for a more institutionalised effort. Yet, we would rather suggest that a dedicated institution may be needed also for effectively fulfilling the initial functions too. 

Such ad hoc AI institution may assume several forms, being either a political body with representatives of states or an expert body, such as the current AI Advisory Body. Likely, a combination of the two could ensure an appropriate level of political support from member states and a strong technical knowledge and expertise.  

While the specific details of this proposed institutions will need a separate discussion, here we would like to highlight three important features. 

First, it should serve the purpose of fostering consensus among member states on AI governance issues, facilitating constructive dialogue on critical topics like climate change and the impact of AI on human rights, and promoting cooperation on AI initiatives aligned with the Sustainable Development Goals (SDGs). 

Second, the proposed AI institution should adopt a multidisciplinary approach by design, leveraging the diverse expertise of various UN agencies to evaluate and promote initiatives and incentives for AI advancements that contribute to achieving the SDGs.

Third, a UN-led global AI institution could be particularly important in the context of AI diplomacy. As the topic of AI increasingly intersects with international relations and diplomacy, the complexities of AI governance require nuanced regulatory frameworks that should come from a super partes entity. Therefore, the proposed AI body should play a pivotal role in putting forward AI model regulations, ensuring that they address the unique challenges and opportunities presented by AI technologies while being adaptable to different countries’ visions of AI. By factoring in the intricacies of AI diplomacy, such as international cooperation and competition, the AI institution can effectively contribute to the development of comprehensive and inclusive AI governance mechanisms on a global scale.

2. An AI institution for fostering AI for SDGs  

The United Nations’ AI institution, given the global support for SDGs and the UN’s role in fostering agreements among member states, should prioritise assessing, devising, and advocating schemes that promote the use of AI to achieve SDGs. One approach is to initiate a dialogue within the World Intellectual Property Organization (WIPO) to explore incentive mechanisms for AI aligned with SDGs. By leveraging intellectual property (IP) as a system of incentives for technology and innovation, stakeholders can gain insights into AI technology and develop mechanisms to foster AI solutions for SDGs.

Intellectual property policies play a critical role in shaping incentives for AI innovation and adoption. The UN AI institution could advocate for IP policies balancing innovation promotion and equitable access to AI technologies for SDGs. Promoting open innovation models like patent pools and open-source licensing arrangements can encourage collaboration among AI developers, particularly for SDG-focused initiatives. Strengthening IP rights protection in developing countries can create an environment conducive to sustainable investment in AI research and development, contributing to sustainable AI innovation globally.

Beyond WIPO engagement, the UN AI institution could collaborate with multinational corporations, academic institutions, and civil society organisations to establish funding mechanisms and grant programs dedicated to AI projects addressing SDGs. Mobilising financial resources and expertise from diverse stakeholders can expedite the development and deployment of AI solutions in areas such as healthcare, education, environmental conservation, and poverty alleviation. Moreover, facilitating knowledge-sharing platforms and capacity-building workshops can empower stakeholders in developing countries to effectively utilise AI for sustainable development.

Therefore, the UN AI institution should focus on incentivizing AI use for SDGs through initiatives such as WIPO dialogues on AI incentives, advocating for balanced IP policies, and collaborating with diverse stakeholders to establish funding mechanisms and capacity-building programs. By integrating intellectual property considerations into the discourse on AI governance for SDGs, the institution can foster a conducive ecosystem for sustainable AI innovation and implementation worldwide, aligning with the UN’s mission of promoting sustainable development and leveraging technology for humanity’s collective benefit.

3. Multi-stakeholder approach by design 

The interim report emphasises, in principle 4, the necessity for universal, networked, and adaptively collaborative governance of AI. To this end, the establishment and design of a dedicated UN institution for AI should embrace a multidisciplinary approach, leveraging the diverse expertise and perspectives inherent in various UN agencies. This entails ensuring direct and robust channels of communication between the AI institution and different UN branches to facilitate informed decision-making and effective coordination. Moreover, principle 5 emphasises the alignment of AI initiatives with the Sustainable Development Goals (SDGs). Therefore, a key objective of the proposed UN AI institution should be to monitor, promote, and provide governance-related advice on AI’s contribution to the SDGs. By prioritising the social benefits of technology and evaluating its environmental and social  impact, such an institution can maximise the positive outcomes of AI for sustainable development on a global scale, and it can use the multi-stakeholder mechanism inspired by guiding principle 4 to achieve such an objective.


In fact, within or connected to the UN, several agencies already play significant roles in the intersection of AI and Sustainable Development Goals (SDGs). These include the International Telecommunication Union (ITU), which focuses on ICT development and standards, and the United Nations Educational, Scientific and Cultural Organization (UNESCO), which promotes the ethical use of AI in the media, education and culture. Additionally, the United Nations Industrial Development Organization (UNIDO) offers insights on integrating AI for industrial development initiatives, while the World Health Organization (WHO) explores AI applications in healthcare for achieving SDG 3. Furthermore, the United Nations Environment Programme (UNEP) addresses AI’s role in environmental sustainability. The International Labour Organization (ILO) can also offer relevant insights on AI and sustainability in relation to labour. The WIPO, in addition to the role suggested in the previous paragraph, can contribute with data on AI innovation and intellectual property-related policies. These agencies’ expertise and mandates contribute to leveraging AI for SDGs across various sectors.

A proposed AI institution within the UN should engage in structured dialogue with these agencies to ensure multi-stakeholder collaboration for AI and SDGs. This collaboration is crucial as it allows for the integration of diverse perspectives and expertise from different sectors. By fostering dialogue and cooperation, the AI institution can leverage the strengths of each agency to develop comprehensive strategies and initiatives for harnessing AI’s potential for sustainable development. Furthermore, involving relevant agencies ensures alignment with existing SDG frameworks and initiatives, maximising synergies and avoiding duplication of efforts. Through structured dialogue, the AI institution can facilitate knowledge sharing, capacity building, and coordination of AI-related activities, ultimately enhancing the effectiveness and impact of AI interventions for achieving the SDGs.

The multi-stakeholder approach of the proposed AI institution should not stop at coordinating efforts within the UN. At regional and national levels, there are several bodies currently working on AI governance or with the potential to do so. From regional human rights systems to national human rights institutions (NHRIs) and domestic regulators, a UN AI central body could develop a structured multi-layered governance network to ensure that the international principles and regulatory guidance are effectively implemented and enforced on the ground. Institutions closer to the individuals and communities that will benefit and suffer from the deployment of AI systems are placed in a privileged position for the effective development and enforcement of AI governance. Yet, they need coordination and legitimacy, which a UN AI body could potentially offer.

The structured dialogue between the proposed AI institution and relevant UN agencies and regional and national bodies could be implemented through a framework that encourages regular communication, collaboration, and coordination. Firstly, establishing formal channels of communication, such as regular meetings, working groups, and joint initiatives, would facilitate information exchange and sharing of best practices. Secondly, incorporating representatives from each agency, regional systems, and NHRIs into the decision-making processes of the AI institution would ensure that diverse perspectives are considered in AI policy development and implementation. Thirdly, developing joint projects and initiatives that leverage the unique strengths and expertise of each entity would enable comprehensive approaches to addressing AI challenges and opportunities for SDGs. Additionally, creating mechanisms for monitoring and evaluation would allow for the tracking of progress, identification of gaps, and adjustment of strategies as needed. Overall, by designing a structured dialogue framework that fosters collaboration and coordination, the proposed AI institution and the UN agencies can effectively harness the potential of AI for advancing SDGs.

4. AI Diplomacy 

The governance and functions of the proposed UN AI institution must address critical aspects of AI diplomacy, including the lack of universally agreed-upon norms and regulations, geopolitical competition, security concerns, ethical dilemmas, the digital divide, and data governance. Diplomatic efforts are essential to foster collaboration, prevent escalation, mitigate security risks, promote ethical standards, bridge inequalities, and manage data flows and privacy issues. 

The complexity of the contemporary geopolitical situation surpasses that of two decades past, marked not only by the emergence of new global powers but also by the proliferation of novel and troubling conflicts that obstacle peace and challenge the enduring relevance and efficacy of the UN. Nevertheless, the UN’s inception in the aftermath of the Second World War was precisely to forestall such scenarios. Thus, it is desirable to reaffirm the commitment to the UN’s foundational mission and mandate during times of adversity. To ensure the UN remains up to date in terms of contemporary dynamics and equipped to face evolving challenges, the establishment of an AI institution holds promise in positioning the UN at the forefront of AI diplomacy. By embracing the transformative potential of AI, the UN can reinforce its pivotal role in promoting global stability, fostering cooperation, and upholding fundamental principles of peace, security, and human rights in the face of new geopolitical complexities.

The discussion presented by Roberts et al. highlights the challenges and considerations surrounding centralised models for AI governance. They address concerns regarding the decentralised and cross-cutting nature of AI policy issues, the risk of regulatory capture by Big Tech companies, and the difficulties in establishing a new international body given the current geopolitical context. As they highlight, addressing challenges related to legitimacy and resource constraints is crucial in this process. Additionally, ensuring inclusive and accountable procedures in AI governance work is also of fundamental importance. In response, they propose strengthening the existing regime complex as an alternative approach, focusing on enhancing coordination, capacities, and legitimacy among existing institutions, drawing parallels with successful models in climate governance. While acknowledging the need to address the challenges, we suggest that the establishment of an AI institution within the UN should not be seen as mutually exclusive to strengthening the existing regime complex but rather complementary actions in addressing the complexities of AI governance.

The UN AI institution could play a crucial role in fostering global consensus on AI principles and strategies, leveraging its capacity for international agreements, potentially resulting in binding treaties. It could gather information, conduct research, and support diverse legislative approaches tailored to different countries’ AI strategies, varying in nuances and granularity.

AI diplomacy could be a primary function of the institution, aligning with the UN’s purposes, including preventive diplomacy to forestall conflicts. While traditional preventive diplomacy focuses on acute crises, AI diplomacy could address the geopolitical challenges posed by AI technologies, adapting diplomatic circles to AI’s economic, security, and ethical impacts. The institution could adopt a preventive AI diplomacy approach, drawing on success factors from previous preventive diplomacy experiences and leveraging its multi-stakeholder framework and multidisciplinary composition. Monitoring AI regulations and governance would allow the institution to identify diplomatic solutions and compromises in line with the UN’s objectives.

Overall, diplomatic engagement, including policy-making, multilateral engagement, and information analysis, is crucial in addressing competitive challenges, legitimacy and reshaping global power dynamics posed by AI technologies. 

FM
Francesca Mazzii
Lecturer in AI, Innovation and Law at Brunel University, London

Francesca Mazzii is a Lecturer in AI, Innovation and Law at Brunel University London. Previously, she was a postdoctoral researcher at Said Business School (Oxford University) on AI and SDGs. She holds a double PhD on patentability of AI-generated inventions in pharma from Queen Mary University London and Maastricht University.

EA
Elena Abrusci
Senior Lecturer in Law at Brunel University, London

Elena Abrusci is a Senior Lecturer in Law at Brunel University London. She served as a Policy Advisor for the UK government on Digital Regulation and researches on AI regulation and the impact of technology on human rights. She's the author of 'Judicial Convergence and Fragmentation in International Human Rights Law (Cambridge University Press, 2023).

[citationic]

Featured Artist