In Évian, on the French shore of Lake Geneva, the G7 leaders did not meet solely to discuss war, economic security, supply chains or the reshaping of the world. They also confirmed the definitive arrival of artificial intelligence in the realm of geopolitics. A few kilometres from Geneva, a city that aspires to become one of the world’s centres of AI governance, the summit sends a signal. Artificial intelligence is no longer a matter of innovation. It is becoming a question of sovereignty, security, power and civilisation.
The symbolism is striking. In 2003, at the previous G7 summit in Évian, the world was still holding its discussions in the shadow of triumphant globalisation, the war on terror and American power. Twenty-three years later, the setting is the same, but the times have changed. States seem less sure of themselves. Alliances are more fragile. Public opinion is gripped by mistrust. And new players from the technology sector are now joining strategic discussions on an equal footing with governments.
The presence around the summit of business leaders from companies such as OpenAI, Google DeepMind, Anthropic, Mistral AI and Cohere speaks to this shift. The major artificial intelligence platforms are no longer merely innovative companies. They own infrastructure, models, computing capacity and vast quantities of data that now bear on national security, industrial competitiveness, the production of information, scientific research and even the conduct of war.
Therein lies the heart of the matter. The governance of artificial intelligence can no longer be conceived of as a technical appendage to digital regulation. It affects the balance of power. It raises an old question in a new language: who controls the instruments of power?
The G7 summit in Évian comes at a particular moment. The United States retains a considerable lead in foundation models, the cloud, semiconductors and computing capacity. Europe, for its part, is trying to turn its anxiety into a strategy. It talks of technological sovereignty, backs its champions and invests in infrastructure, yet remains largely dependent on industrial and technological ecosystems that are not its own. As for the countries of the Global South, they risk once again finding themselves in the position of consumers of standards, systems and architectures designed elsewhere.
That is why the debate on artificial intelligence goes far beyond the question of innovation alone. It concerns strategic dependence. It concerns the ability of states to preserve their autonomy in decision-making. It concerns the place of societies in defining the rules that will govern technologies capable of affecting employment, education, health, justice, security and information.
The G7 can open up a path. But it cannot claim, on its own, to define the global order for artificial intelligence. Therein lies the whole ambiguity of this summit. The advanced industrial democracies are right to want to contain the risks, establish safety principles, build trust and prevent the most powerful technologies from becoming uncontrolled instruments. But they cannot do so within a closed circle, among a handful of states and a handful of dominant companies.
AI governance will be legitimate only if it is inclusive. It will have to involve the technological powers, states, international organisations, researchers, businesses and civil society, but also emerging countries. For the effects of artificial intelligence will not stop at the borders of the G7. They will reach African administrations, Asian education systems, Arab cities, Latin American economies, fragile health systems, independent media and societies already vulnerable to information manipulation.
This is where Geneva has a card to play.
A few kilometres from Évian, International Geneva is following this debate closely. It knows that the governance of artificial intelligence will not be built at political summits alone. It will also be built in the forums where states, specialised agencies, NGOs, diplomats, experts, industry and representatives of civil society come together. Geneva has this culture of technical and political dialogue. It has experience in standards, multilateralism, humanitarian affairs, human rights, trade, health and telecommunications. In other words, it has a rare ecosystem for addressing AI not as a mere technology, but as a global phenomenon.
The timing reinforces this impression. A few days after the G7 summit in Évian, Geneva is due to host the first United Nations Global Dialogue on Artificial Intelligence Governance, alongside AI for Good and the World Summit on the Information Society Forum. The city thus finds itself at the heart of a decisive international sequence. On one side, Évian enshrines AI as a question of power. On the other, Geneva can attempt to bring it back within the realm of rules, accountability and the general interest.
This geographical proximity is more than a coincidence. It tells the story of the tension of our times. In Évian, the most powerful states discuss sovereignty, security and technological dominance. In Geneva, multilateralism is still seeking to create spaces for trust and cooperation. Between the two shores of Lake Geneva, two visions of global governance answer each other: that of the club of powers and that of the universal forum.
The point is not to set them mechanically against each other. The G7 has its uses. It allows rapid progress among countries with comparable interests. But it cannot be enough. Artificial intelligence calls for broader, more transparent and more balanced governance mechanisms. Safety standards, audits, certifications, interoperability mechanisms, rules on access to advanced models and safeguards against misuse will have to be discussed in open formats. Otherwise, AI governance risks itself becoming an instrument of domination.
The risk is obvious: the emergence of a multi-speed world. On one side, countries able to build, train and control the large models. On the other, states dependent on imported technologies, unable to audit their effects, understand their biases or negotiate the terms of their use. Such a divide would not be merely digital. It would be political. It would create a new international hierarchy of decision-making, knowledge and influence.
The battle over AI governance is therefore also a battle for trust. Trust in systems. Trust in data. Trust in institutions. Trust in the ability of societies not to be stripped of their democratic choices by opaque technologies. At a time when disinformation, deepfakes, influence operations and automated content are already weakening the public sphere, artificial intelligence cannot be approached solely from the standpoint of productivity.
It compels us to think about the relationship between technology and democracy. It forces us to ask who decides, who controls, who certifies, who is accountable for errors, who protects citizens and who guarantees that innovation does not turn into a collective vulnerability.
It is precisely on this ground that Geneva can set itself apart. Not by claiming to compete with Silicon Valley, Shenzhen, Paris, London or Toronto on the industrial front, but by offering a place for mediation, certification, normative dialogue and trust. The Geneva of AI will not necessarily be home to the largest models. It can be the Geneva of rules, standards, independent evaluation, interoperability and accountability.
That ambition, however, must be pursued with clarity. Geneva cannot be content merely to host conferences. It must build a genuine political and institutional offer around AI. It must bring together its international organisations, its universities, its foundations, its private-sector players, its diplomats and its experts. It must become a place where the countries of the Global South also have a seat at the table. For global AI governance without Africa, without emerging Asia, without the Arab world and without Latin America would be merely partial governance, and therefore fragile.
Finally, the G7 summit in Évian is a reminder of a harsh truth: states want to govern artificial intelligence at the very moment they are discovering that they are no longer the only ones producing power. Technology companies are not replacing states, but they are constraining them. They possess capabilities that governments do not always master. They move faster than diplomatic procedures. They sometimes impose their standards before regulators have even had time to discuss them.
That is the whole challenge of the years ahead: preventing AI governance from being written solely by those who own the machines, the data and the infrastructure.
In Évian, the G7 leaders opened a new chapter. In Geneva, the task will be to try to give it multilateral depth. The first summit speaks of power. The second city can serve as a reminder of the rules. Between the two, Europe and Switzerland bear a particular responsibility: to make artificial intelligence not merely a matter of competition, but a matter of trust, law and cooperation.
For the real question is no longer whether AI will transform the world. It is already doing so. The question is whether the world will be able to govern this transformation before it ends up doing the governing instead.
GAIGI, a Geneva initiative serving the global governance of AI
Founded in Geneva at the initiative of Axel Mazolo, the Geneva AI Governance Institute (GAIGI) is part of the drive to make Geneva one of the world’s centres for the governance of artificial intelligence. The foundation carries out work on regulatory interoperability, certification, digital trust and the oversight of so-called “sensitive” AI systems, particularly in the fields of health, defence, critical infrastructure and diplomacy. (Geneva AI Governance Institute)

At a time when the major powers are debating technological sovereignty and AI safety in Évian, GAIGI advocates a complementary approach based on multilateral dialogue, the inclusion of the Global South and the building of international trust mechanisms. Based at the heart of International Geneva, the institution works with international organisations, states, businesses and civil society actors to help bring about norms and governance frameworks suited to the challenges of the 21st century.
The ambition is clear: to make Geneva not only a venue for debate on artificial intelligence, but also a benchmark hub for developing the standards, certification mechanisms and trust architectures that will accompany the global development of AI. This vision ties in with Geneva’s growing stature as an international platform for dialogue on AI, illustrated by its hosting of the United Nations Global Dialogue on AI Governance and the AI for Good summit. (ITU)
Translated from French. The French original prevails.





