A Broken World Order Leaves AI as Multilateralism's Last Test

A Broken World Order Leaves AI as Multilateralism's Last Test

Feb 7 , 2026. By Margrethe Vestager ( Margrethe Vestager, a former vice president of the European Commission and EU commissioner for competition, is co-president of the Europa Power Initiative and chair of the Board of the Technical University of Denmark. This article is provided by Project Syndicate (PS). )

The world order’s troubles predate the Trump presidency. The UN has struggled to function effectively, and the WTO has lost momentum. Middle powers such as India, Brazil, and South Africa have openly questioned the legitimacy of international systems anchored in Western priorities. As institutions lose their legitimacy, countries increasingly pursue unilateral interests, often leading to trade disputes and unresolved conflicts.

When people ask me how I am doing, I generally reply: "I'm fine, but the world is a mess." Still, as a Dane, the past few weeks have been particularly difficult, and incomparably worse for the people of Greenland.

With his assertion that might-makes-right, his threats to Danish sovereignty, his undermining of the United Nations (UN) through a Board of Peace, and his commercialisation of humanitarian aid, President Donald Trump has made his worldview plain, and it is deeply troubling.

But the uncomfortable truth is that the world order was already broken before Trump's first presidency. He did not create the dysfunction. He simply poured gasoline on the fire and accelerated the decline. The UN had long failed to function effectively, and the World Trade Organisation (WTO) had ground nearly to a halt. Major regional powers such as India, Brazil, and South Africa were openly questioning the legitimacy of an international system that remained wedded to a Western perspective and too often failed to reflect their perspectives or accommodate their interests.

The real danger now is that Trump-incited chaos will be used as an excuse for paralysis; that we will be so busy defending the old order that we fail to build a better one. It is all too tempting to circle the wagons, to defend existing institutions as a matter of principle or out of a sense of duty. But this defensive approach misses the essential point. The alternative to a dysfunctional order is not the same order functioning better. It is a better order, or no order at all.

When international institutions lack legitimacy, countries pursue their interests unilaterally. When the WTO cannot resolve disputes, governments eventually resort to tariff wars. When the UN Security Council is paralysed, conflicts metastasise, and the costs fall most heavily on smaller countries and the global commons. We have seen this time and again when it comes to tackling climate change, pandemics, cybersecurity, and other collective challenges.

Trump's might-makes-right approach only works because our institutional constraints have already failed. Now that he is gleefully destroying what remains of the old order, a strategy for reform and renewal is not a luxury that we can afford to defer. The erosion of institutional legitimacy creates the very conditions that allow leaders like Trump to thrive.

For proof that reforming global institutions is no longer an option, look no further than the challenges posed by AI. These technologies, with their extraordinary potential benefits and equally extraordinary risks, cannot be governed by any single country, no matter how powerful it is. Effective governance requires exactly what we are missing: legitimate and effective global cooperation.

That means we have an opportunity. Unlike reforming institutions burdened by decades of accumulated dysfunction and resentment, we can build governance frameworks for AI from the ground up, and what we build can properly reflect today's multipolar reality, rather than yesterday's Western dominance.

The Hiroshima AI Process, launched under Japan's G7 presidency in 2023, offers a model that we can build on. It brought together major economies to establish voluntary guidelines for AI development and deployment. But, of course, voluntary guidelines among a limited group of countries are not enough. What we really need is a global framework that includes the Global South, that strikes an optimal balance between innovation and safety, and that has functioning enforcement mechanisms.

This is not about creating a new bureaucracy. It is about establishing clear principles on safety, transparency, legal liability, and the rights of affected populations. That is how all countries can be confident that AI will develop in ways that serve humanity, rather than narrow national or private interests.

The comparison to nuclear weapons is instructive, even if it is imperfect. AI will not be contained by non-proliferation treaties, because the technology is simply too distributed for that. Instead, we need something closer to the frameworks governing aviation safety or pandemic surveillance. These function through technical cooperation grounded in shared self-interest, with mechanisms for rapid information sharing and coordinated responses to emerging risks.

We need frameworks that are inclusive, practical, and genuinely empowering. Establishing effective AI governance could demonstrate what reformed multilateralism looks like and restore confidence that international cooperation delivers real value. And it could establish a template for tackling other challenges that do not stop at national borders.

The world is indeed a mess. But defending institutions that have lost their effectiveness and legitimacy is not the answer. We can and should build something better, starting with AI. The alternative is not preserving the status quo. It is watching it collapse entirely.

PUBLISHED ON Feb 07,2026 [ VOL 26 , NO 1345]