How Global AI Governance Can Build a Better World

Artificial intelligence and other new technologies are advancing at an unprecedented rate, reshaping how people live, learn, and work. Amid this rapid transformation, governments, companies, and organizations are grappling with a fundamental question: How can the world maximize and share the benefits of AI while managing its risks?

This question will dominate the agenda when countries, civil society, and the private sector come together for the UN’s first Global Dialogue on AI Governance, one of the most anticipated gatherings on artificial intelligence in Geneva on July 6–7, 2026.

I sat down with the UN Foundation’s Vice President for AI and Digital Cooperation Strategy, Dr. Claire Melamed, to discuss the debates, myths, and imperatives around AI governance.

This conversation has been edited for clarity and length.

AI GOVERNANCE, EXPLAINED

Christina: What is AI governance, and why should people care about it?

Claire: AI seems to be everywhere, and people are becoming increasingly worried about it. They worry that AI might cost them their jobs, and there are growing concerns in many countries about everything we do not yet know about what AI could do.

AI governance is really about society saying: What is it that we want AI to do for us? And how can we set the terms so that we get good outcomes from AI, and it doesn’t harm us?

In democratic countries, we look to governments — as the representatives of society — to be the ones that, through consultations with civil society, the private sector, and others, set those terms. What AI governance should be is a set of decisions that governments take on behalf of their populations to make sure that AI works for them and not against them. That sounds simple and, of course, it’s more complicated than that, but that’s the basis of what governments should be thinking about.

Christina: In a few weeks, the UN will hold the first-ever Global Dialogue on AI Governance. What is the Global Dialogue, and why is the UN uniquely positioned to help shape global conversations on AI governance?

Claire: It seems like policymakers are talking about AI every time they get together. However, most of those conversations involve just a few governments, such as the richer countries that meet through the G7. A lot of countries are not really involved in any of those discussions, but their populations will be as affected by AI, so they need to have a say. The Global Dialogue invites all governments to participate, and they are also inviting the private sector, civil society, and others to join.

There are two reasons the UN is the place for these conversations. The first is inclusiveness. The UN is where all governments can come together and talk about this global — as well as local — issue. It is the only place where everybody is equal.

If AI is going to have the far-reaching consequences for societies, economies, culture, and language that many of its makers believe it will, it is going to affect everybody, whether you’re a tech worker in Silicon Valley or a farmer in the Sahel. The UN is the only place where all those players can come together.

The second reason is more structural. During a big global crisis — whether it’s climate change, nuclear weapons, or a pandemic –– governments turn to the UN. When there are new global coordination challenges from rules for international aviation to shipping, the UN is the place for that, too. Let us hope we never reach a stage where we see AI as a problem on par with nuclear weapons. But it’s a plausible scenario, and whether it comes to that or not, we are going to want the UN to be involved. That is why it is better to lay down the infrastructure now, so that the UN is ready when that moment comes.

For this first Global Dialogue, the focus is really on setting out the terms of engagement: What are the aspects of the AI conversation that the UN is the right forum for, in addition to everything else already going on? What the UN can uniquely do is find the points where different regional and national discussions about AI intersect, and where the whole world needs to come together. If we can answer those questions and set an agenda for future UN conversations, I will consider the Global Dialogue a success.

DEBUNKING THE MYTHS AROUND AI GOVERNANCE

Christina: Some critics argue that global governance of AI could slow innovation. How do you respond to that concern?

Claire: Innovation is not a single, linear thing that can only happen in a certain way. Innovation always responds to the constraints placed upon it, and sometimes the limits come from the technology itself. For example, the AI innovation we have seen recently has been driven on scale with more data centers and bigger capacity leading to more progress. Some people think we are starting to hit the limits of that approach, both in material terms and as a broader strategy. If that turns out to be the case, then we will have to find a way around those limits. Often, that’s where innovation generates the biggest breakthroughs.

The same is true for constraints imposed by governments. The idea that innovation is some fragile thing that could be blocked by laws is simply not borne out by any historical example. When we introduced standards for car engine emissions, for example, it spurred a whole wave of innovation in how to produce cleaner engines. Innovation responds to whatever the context is.

If government regulation creates incentives — to do more of this and do less of that — the industry will adapt, and we will get innovation in those areas. That is the whole point of government policies: to change behavior. We should not worry that the kind of AI governance being discussed will crush innovation.

A GLOBAL FORUM FOR THE AI DEBATE

Christina: How can the Global Dialogue help ensure more inclusive participation in AI governance?

Claire: It is important to recognize the extent to which the trajectories of AI are being shaped by individual decisions taken in very few places. This is not sustainable as AI becomes a global industry, and even some of the companies themselves are starting to call for a global approach and to act responsibly given the lack of rules. For example, Anthropic is a company that has put safety at the heart of its decision-making. They recently had a very powerful new model and chose not to make it publicly available because of concerns about what it could do. Instead, they used a careful, staged [phased] process and released it to a small number of companies and one or two governments to explore it, identify risks, and consider how those risks could be mitigated.

That was the right thing to do. But Anthropic did not have to do it, and the next company to build a powerful product might not make the same responsible choices. It is alarming that how the public experiences AI — and whether we are exposed to certain risks — essentially comes down to decisions made in the executive offices of one or two companies. That is not a situation that fills me with confidence, and it’s why we need some way to discuss these issues with everyone around the table. The Global Dialogue is the first step.

BUILDING A PEOPLE-FIRST AI FUTURE

Christina: How can we make sure AI does not widen the gap between the world’s richest and poorest countries?

Claire: All technologies, especially at the beginning, are distributed somewhat unequally; there is always some degree of diffusion over time. The risk is that those differences get bigger instead of smaller, and then they start to become a new kind of structural inequality. That is what we need to try to prevent while we still have time.

One of the essential principles of the UN is sustainable development. The job here is not just to make sure that AI does good, but to make sure that AI gets to the countries that need it as quickly as possible.

The development agenda has a number of components. First of all, it’s about investment and laying down the infrastructure: the cables, the data centers, and the hardware.

We also need to make sure that communities in need of this technology know how to use it to build solutions and respond to their wants. It is about making sure that companies –– whether in Kenya, in Mali, or in Timor-Leste –– can build what communities need and can get the finance, support, and connections to trading systems so they can succeed.

These challenges cannot all be solved at once, and they require different kinds of relationships and stakeholders. But the UN has always been the main global convenor and catalyzer of this kind of partnership, and it is no less important in regard to AI.

Christina: Looking ahead, what would responsible and effective AI governance look like?

Claire: Like all policymaking, AI governance is a series of trade-offs. Governance is not the art of everyone getting what they want — it is the art of managing different priorities, interests, and needs. With something as fast-moving and difficult to grasp as AI, people do not always know what they want now, let alone in five or 10 years. Ultimately, the test must be whether people feel happier and more secure.

On one side of the ledger, there is a catalog of enormous potential benefits. Will there be innovations in healthcare? Will people’s working lives become more productive and more satisfying? Will we find new ways to tackle climate change? Good governance can speed up the realization of those benefits and ensure their distribution across all countries and communities.

On the other side, there are real risks such as existential concerns about AI getting out of control. There are also more intermediate risks. For example, will AI distort how information is produced and shared so we stop being able to believe anything? If that happens, how does democracy work? How do we maintain a shared sense of trust in justice systems, in education, and in institutions that have underpinned social stability?

None of this is hard-wired into the technology itself. AI technology is going to have certain impacts if we make certain choices and different impacts if we make different choices. No one should accept any narrative that suggests otherwise.

Because so much of the debate about AI has been dominated by those at the cutting edge of making it, we risk confusing two very different things: 1) What technologists say about the capabilities of the technology, and 2) The separate set of choices that individuals, people, and institutions will need to make about how we relate to that technology. There is absolutely nothing inevitable about that second set of choices.