-Alina Vey

Artificial intelligence is increasingly being discussed as more than a technological development. It is becoming a question of international governance, economic power, security and the future relationship between states.

The technology itself is developing across borders. Research is international, companies operate across jurisdictions, data moves between countries, and the consequences of increasingly capable systems will not necessarily remain within the country in which those systems are developed.

Yet governance remains largely organised around national jurisdictions.

This creates a fundamental question:

What happens when the technology is global, but many of the rules governing it remain national?

The question is becoming increasingly difficult to ignore.

In September 2026, the United Nations General Assembly President addressed the issue of multilateral AI governance, arguing that international discussions must address common understandings around areas including safety, accountability, human oversight, interoperability and access to AI capabilities. The United Nations has also begun developing a broader international dialogue around the governance of artificial intelligence.

At the same time, the House of Lords International Relations and Defence Committee has been examining the effectiveness of existing multilateral structures for AI governance, including the ability of countries such as the UK to contribute to the development of international rules and norms.

And on 21 September, OpenAI called for international technical standards for advanced AI, including shared approaches to evaluating capabilities and reporting serious incidents.

These developments point towards a problem that extends beyond the question of how AI should be regulated within individual countries.

They raise the question of whether international governance can develop quickly enough to keep pace with the technology itself.

A technology without a single border

Previous technological transformations have often been associated with particular industries, countries or physical infrastructures.

AI is different in an important respect.

Its development depends upon networks of researchers, companies, computing infrastructure, investment, data and international supply chains. Its applications can then spread rapidly across borders.

An AI system developed in one country can be used in another. A cyber capability can cross jurisdictions. An automated system can influence financial, commercial or information environments far beyond the territory in which it was created.

This does not mean that every consequence of AI is inherently international. Much of its regulation will necessarily remain national or regional.

But it does mean that purely national approaches have limits.

A government can establish rules for companies operating within its jurisdiction. It can establish requirements for particular sectors. It can determine how public institutions use AI.

It cannot, by itself, determine how every other country develops, deploys or governs the technology.

That is where the international dimension begins.

Competition and cooperation

There is an understandable tension between technological competition and international cooperation.

Artificial intelligence is increasingly connected with economic competitiveness, scientific research, national security and strategic influence. Governments therefore have strong reasons to develop their own capabilities and protect their technological advantages.

Competition between states is unlikely to disappear simply because countries recognise the need for cooperation.

The more realistic question is whether the two can exist simultaneously.

Historically, states have cooperated in areas where they also compete. International aviation, telecommunications, nuclear safety, financial systems and many other areas have required some degree of shared rules even when the countries involved have had different interests.

AI presents a similar problem, although its speed of development and breadth of application create particular difficulties.

Countries may compete to develop the most capable systems while still having a shared interest in establishing standards for safety, testing, transparency, security and the reporting of serious failures.

Cooperation does not necessarily require strategic convergence.

It can instead provide mechanisms for managing areas where national interests overlap.

Who gets to write the rules?

There is another question that is perhaps less frequently discussed.

Who gets to shape the international rules?

Governance is not simply a technical exercise. The standards established today can influence which technologies are developed, how they are deployed, who has access to them and how their risks are assessed.

If international standards emerge, the countries and institutions involved in developing them will inevitably have greater influence over the technological environment that follows.

This creates an important geopolitical dimension.

The issue is not simply whether countries agree that AI requires governance.

It is also whether they agree on what that governance should look like, which institutions should provide it, whose interests should be represented and how disagreements should be resolved.

The United Nations has emphasised the importance of ensuring that countries have a voice in shaping the development of AI. Its General Assembly President has pointed to the risk of fragmented systems of governance widening existing divisions and leaving countries with fewer technological capabilities further behind.

This raises a wider issue of international legitimacy.

If the most consequential decisions about advanced technologies are shaped primarily by a relatively small number of technologically advanced states and companies, other countries may find themselves governed by standards in whose creation they had limited influence.

That does not mean that every country must have identical technological capabilities.

It does mean that the question of representation cannot be separated entirely from the question of governance.

The danger of fragmentation

One possible future is not the emergence of a single global system of AI governance, but several competing systems.

Different countries could establish different standards for safety, transparency, data, testing and deployment.

Companies operating internationally could then face increasingly complex regulatory environments.

More importantly, divergent standards could create gaps between jurisdictions.

Technology does not necessarily become safer simply because there are more rules. The effectiveness of governance depends partly upon whether those rules can operate across the environments in which the technology is actually being developed and used.

This is one reason interoperability and international coordination have become increasingly prominent in discussions about AI governance.

Yet coordination also has limits.

International agreements can be slow to negotiate. Countries have different economic interests, security priorities, political systems and technological capabilities.

There is therefore no guarantee that a single comprehensive global framework will emerge.

The more realistic possibility may be a combination of national regulation, regional frameworks, international standards and cooperation between governments, technology companies and scientific institutions.

The challenge will be ensuring that these different layers do not become so disconnected that they undermine one another.

The role of technology companies

There is another unusual feature of the AI era.

Some of the most advanced AI systems are being developed by private companies rather than governments.

That creates a different relationship between technological capability and political authority.

Companies can possess enormous technical expertise and resources, while governments retain responsibility for legislation, national security and public accountability.

Neither side can simply replace the other.

Governments require technical knowledge to create effective policy. Companies operate within legal and political systems that establish the boundaries within which their technologies can be developed and deployed.

The question, therefore, is not whether governments or companies should control AI in isolation.

It is how appropriate responsibilities can be divided while preserving democratic oversight and public accountability.

Recent proposals for international AI standards reflect this tension. OpenAI has argued that national governance should be connected through international cooperation and that shared standards are ultimately necessary because advanced AI systems, research and expertise move across borders.

That is a significant development in itself.

The debate over AI governance is moving from a question of whether regulation is necessary towards a more difficult question:

What kind of governance is actually capable of operating at the scale of the technology?

Beyond regulation

It is tempting to describe the challenge simply as one of regulation.

But regulation is only part of the picture.

The future of AI will also depend upon access to computing infrastructure, research, education, data, investment and technical expertise.

Countries with limited technological capacity may become increasingly dependent upon technologies developed elsewhere.

That creates questions about economic inequality and technological sovereignty alongside questions of safety.

If AI becomes deeply embedded in healthcare, education, defence, finance, communications and public administration, technological capability will become increasingly important to national resilience.

The distribution of that capability could therefore influence international relations as much as the technology itself.

This makes international cooperation more complicated.

Countries may want to share knowledge and establish common standards while simultaneously trying to protect strategic technologies and maintain competitive advantages.

That tension is unlikely to disappear.

It will have to be managed.

A new test for international cooperation

Artificial intelligence therefore presents a broader test of the international system.

The question is not whether countries will compete.

They will.

Nor is it whether governments will regulate AI.

They increasingly are.

The deeper question is whether international institutions and national governments can develop forms of cooperation capable of operating alongside technological competition.

If they can, AI could become an area in which technological advancement is accompanied by stronger international mechanisms for managing shared risks and widening access to its benefits.

If they cannot, the alternative may be a more fragmented technological landscape in which different standards, competing systems and unequal access reinforce existing geopolitical divisions.

Neither outcome is predetermined.

The direction will depend upon choices made by governments, institutions, companies and societies over the coming years.

What is increasingly clear, however, is that artificial intelligence is no longer simply a question for the technology sector.

It is becoming part of the wider structure of international relations.

And perhaps the most important question is no longer simply what AI will become.

It is who will have a voice in deciding what it becomes — and how the international system will govern it when it does.

Alina Vey