Global AI governance should be anchored by five key principles

Policymaking is moving almost as fast, with the launch of new regulatory initiatives and fora seeking to meet the moment. But while ongoing efforts by the G7, EU and the US are encouraging, none of them is universal, representing the global commons. In fact, with AI development driven by a handful of CEOs and market actors in just a few countries, the voice of the Global South has been absent. The unique challenges that AI poses demand a coordinated global approach to governance, however, and only one institution has the inclusive legitimacy needed to organize such a response: the United Nations. We must get AI governance right if we are to harness its potential and mitigate its risks. With that in mind, the UN High-level Advisory Body on AI was established to offer analysis and recommendations for addressing the global governance deficit. It has 38 individuals from around the world, representing diverse geographies, genders, ages and disciplines, and drawing on expertise from government, civil society, academia and the private sector.

Privileged to serve as the Advisory Body’s Executive Committee, we have released the group’s interim report, which proposes five principles for anchoring AI governance and addressing several inter-related challenges.

First, since risks differ across diverse global contexts, each will require solutions that are tailored accordingly. But that means recognizing how rights and freedoms can be jeopardized by specific design, use (and misuse) and governance choices. Failing to apply AI constructively (“missed uses”) can needlessly exacerbate existing problems and inequalities.

Second, since AI is a tool for economic, scientific and social development, and since it is already assisting people in daily life, it must be governed in the public interest. So we must bear in mind goals related to equity, sustainability and societal and individual well-being, as well as broader structural issues like competitive markets and healthy innovation ecosystems. Third, the emerging regulatory frameworks across different regions will need to be harmonized in order to address AI’s global governance challenges effectively. Fourth, AI governance should go hand-in-hand with measures to uphold agency and protect privacy and the security of personal data. Lastly, governance should be anchored in the UN Charter, international human-rights law, and other international commitments where there is a broad global consensus, such as the UN Sustainable Development Goals.

Affirming these principles in the context of AI requires overcoming stubborn challenges. AI is built on massive amounts of computing power, data and specific human talents. Global governance must consider how to develop and ensure broad access to all three. It also must address capacity building for the basic infrastructure that underpins the AI ecosystem (such as reliable broadband and electricity), especially for the Global South. Greater efforts also are needed to confront both known and still-unknowable risks that could emerge from AI’s development, deployment or use. AI risk is a hotly debated subject. While some focus on eventual end-of-humanity scenarios, others are more worried about harms to people here and now; but there is little disagreement that the risks of ungoverned AI are unacceptable.

For good governance, we foresee the need for objective assessments of the state of AI and its trajectory, to give citizens and governments a sound foundation for policy and regulation. At the same time, an analytical observatory to assess AI’s societal impact (from job displacement to national-security threats) would help policymakers keep up with the changes that AI is driving offline. The world will need the capacity to police itself, including by monitoring and responding to potentially destabilizing incidents and by facilitating accountability and enforcement action.

These are just a few of the recommendations we are advancing. They should be seen as a floor, not a ceiling. More than anything, they are an invitation for more people to tell us what kinds of AI governance they would like to see.

If AI is to fulfil its global potential, new structures and guard-rails are needed to help us all thrive as it evolves. Everyone has a stake in AI’s safe, equitable and accountable development. The risks of inaction are also clear. We believe that global AI governance is essential to reap the opportunities and navigate the risks that this technology presents every state, community and individual today and for generations to come. ©2024/project syndicate

James Manyika and Marietje Schaake are co-authors of this article.