News

AI Governance Is Becoming Infrastructure

• By John K. Waters
• 07/16/2026

For the past three years, the conversation around AI governance has centered on rules. Governments debated regulations. Companies signed voluntary commitments. Researchers warned about frontier models. International summits produced declarations, communiqués, and promises to keep talking.

This month's AI for Good Global Summit in Geneva suggested something different is now underway. The biggest story out of the summit was not a new model, a benchmark, or a new regulation. It was the emergence of institutions.

Alongside the summit, the United Nations convened the inaugural Global Dialogue on AI Governance, an ongoing forum established by the UN General Assembly to bring governments, industry, researchers, and civil society together to discuss international AI governance. The International Telecommunication Union also launched the AI for Good Global Commission, a permanent body co-chaired by Rwandan President Paul Kagame and Salesforce Chair and CEO Marc Benioff to help guide international AI initiatives.

Neither organization has regulatory authority. Neither writes laws. Neither can compel companies or governments to do anything. What they do instead is create a place where governments, companies, researchers, and standards bodies can meet repeatedly over time. That matters because most international standards and norms do not begin as laws. They begin as conversations, working groups, draft frameworks, and technical recommendations that eventually harden into consensus.

From policies to institutions
Technology industries rarely mature through technology alone. They mature when institutions emerge around them. Civil aviation has the International Civil Aviation Organization. Global telecommunications have the ITU. Nuclear energy is overseen by the International Atomic Energy Agency. International trade is governed by the World Trade Organization. Each provides a venue where governments, experts, and industry can develop standards, exchange information, coordinate policies, and build trust, even when they disagree on specific regulations.

Artificial intelligence has spent the past several years in an earlier phase, defined by national AI strategies, executive orders, voluntary safety commitments, and high-profile meetings such as the AI Safety Summits held in the United Kingdom, South Korea, and France. Those efforts remain important, but they have largely been episodic.

The initiatives announced in Geneva suggest international organizations are beginning to treat AI governance as a continuous process rather than a series of individual events.

Infrastructure is more than hardware
The word "infrastructure" usually brings to mind data centers, fiber networks, electrical grids, and increasingly, GPUs. That is certainly part of the story: Microsoft, Amazon Web Services, Google Cloud, Oracle, Nvidia, and others continue to invest billions of dollars in the computing capacity needed to train and deploy increasingly capable AI systems.

But infrastructure is also institutional. It includes the organizations that establish technical standards, the forums where governments negotiate common approaches, the processes that let countries exchange information, and the mechanisms that help smaller nations participate in technologies largely developed elsewhere.

Several sessions at AI for Good reflected that broader definition. The agenda devoted significant attention not only to frontier AI capabilities but also to computing access, AI standards, digital inclusion, healthcare, environmental sustainability, robotics, and multilingual AI development, recognizing that AI's benefits will depend as much on broader participation in the ecosystem as on more capable models.

The next phase of AI
None of this means global AI regulation is imminent. The world's major economies continue to pursue different approaches, and those differences are unlikely to disappear: the European Union is implementing the AI Act, the United States has emphasized sector-specific oversight and voluntary commitments, and China continues to develop its own regulatory framework.

The UN's new dialogue is not intended to replace national policy. Its purpose is to provide a recurring venue where those conversations can continue despite political differences, an outcome that may ultimately prove more durable than another declaration signed at the end of a conference.

Why enterprises should pay attention
It is easy to dismiss international forums as distant from day-to-day enterprise IT, but that would be a mistake. Many of the technical standards businesses rely on today emerged from years of collaboration among governments, standards bodies, and industry groups before becoming procurement requirements, compliance frameworks, or accepted best practices.

AI is likely to follow a similar path.

Organizations deploying AI today are understandably focused on models, agents, security, governance, and return on investment, and those priorities are not going away. But beneath them, another layer is taking shape: international governance forums, shared technical standards, common terminology, interoperability frameworks, and mechanisms for cross-border cooperation.

None of it will make headlines the way a new frontier model does, yet history suggests that this kind of groundwork often shapes industries more profoundly than individual technology announcements do.

The biggest takeaway from Geneva may not be that the world agreed on how to govern AI. It clearly has not. The more important development is that the world is beginning to build the institutions where that conversation will continue. If the first phase of the AI era was about creating increasingly powerful models, the next phase may be about building the infrastructure needed to govern, deploy, and sustain them.

Why it matters
The AI industry has spent much of the past three years focused on model capability, benchmark scores, and competitive positioning. The developments in Geneva point to a parallel trend receiving far less attention: the emergence of permanent institutions designed to support international cooperation around AI. Whether or not they ultimately influence regulation, these organizations are likely to become part of the governance infrastructure surrounding enterprise AI for years to come.