Bahar Serçin Halefoğlu

The reflections below are shaped by a series of recent international dialogues, high-level policy discussions and expert exchanges on AI governance and diplomacy. Rather than focusing on a single event, they seek to capture a broader shift that appears to be taking place in global AI governance.

For much of the last decade, global conversations on artificial intelligence have revolved around a familiar set of concepts. Transparency. Fairness. Accountability. Human oversight. Explainability. Safety.

These principles have become the common vocabulary of AI governance. They appear in national AI strategies, multilateral declarations, ethical guidelines and international policy discussions across the world.

Yet, after following recent international dialogues on artificial intelligence, I have come to believe that something more fundamental is beginning to emerge. The debate is gradually moving beyond what responsible AI should look like. Instead, it is beginning to ask a different question:

What kind of international governance architecture will be capable of governing AI itself?

This distinction may appear subtle, but it represents an important shift. For years, discussions largely focused on defining responsible behaviour for AI developers and users. Today, however, the discussion increasingly concerns the institutions, governance models and diplomatic mechanisms that will coordinate AI across countries, sectors and political systems.

In other words, AI governance is itself becoming an object of governance.

One of the most striking observations from recent international discussions is the remarkable level of convergence around fundamental objectives. Regardless of political systems, economic priorities or geographical regions, a number of themes repeatedly appear across global conversations.

Building trustworthy AI. Ensuring meaningful human oversight. Protecting fundamental rights. Strengthening international cooperation. Supporting capacity building. Reducing global inequalities in access to AI technologies.

While disagreements certainly remain, particularly regarding implementation, the broader vision appears increasingly shared. This represents a significant development.

Only a few years ago, discussions around AI governance often seemed fragmented, with countries emphasizing different priorities and speaking through very different policy languages. Today, despite continuing geopolitical competition, a common governance vocabulary is beginning to take shape. This should not be underestimated.

Shared language often precedes shared institutions.

However, convergence around principles should not be confused with convergence around governance. If anything, recent developments suggest the opposite. Rather than moving toward a single global governance framework, the international landscape is becoming increasingly diverse.

Multiple institutions are developing governance models. Multiple expert communities are producing recommendations. Multiple regulatory approaches are evolving simultaneously. Technical standards continue to mature. Public procurement requirements are becoming governance instruments in their own right.

Research organizations, governments, international organizations and industry are all contributing different pieces of an emerging governance ecosystem. This diversity is often described as fragmentation. I am not convinced that this is necessarily the right interpretation.

At this stage of technological development, institutional diversity may actually be a sign of healthy adaptation. Artificial intelligence is evolving faster than almost any previous general-purpose technology. Expecting one institution — or one governance model — to anticipate every challenge may neither be realistic nor desirable.

Different organizations possess different expertise. Some contribute scientific assessment. Others develop technical standards. Some focus on regulatory coordination. Others specialize in implementation, education or capacity development.

Viewed from this perspective, today's governance landscape resembles less a fragmented system than an evolving ecosystem.

The real challenge, therefore, may not be institutional diversity. It may be institutional interoperability. As more governance initiatives emerge, their ability to communicate, cooperate and reinforce one another becomes increasingly important.

Interoperability should not be understood only as a technical concept. Just as digital systems require interoperable standards, governance systems also require compatible principles, compatible terminology and compatible institutional processes.

Future AI governance will likely depend upon convergence across multiple dimensions simultaneously: normative convergence around shared values, regulatory convergence where appropriate, technical convergence through internationally recognized standards, and operational convergence through procurement frameworks, auditing mechanisms and implementation practices. Without these different layers working together, even well-designed governance initiatives risk operating in isolation.

The question is therefore no longer simply whether AI should be governed. It is whether governance systems themselves can become sufficiently coordinated to govern AI effectively.

This evolution also has profound implications for diplomacy. Artificial intelligence is often discussed as another policy issue alongside climate change, cybersecurity or digital trade. Increasingly, however, AI appears to occupy a different position.

Climate diplomacy focuses on climate. Cyber diplomacy focuses on cyberspace. Trade diplomacy focuses on markets. Artificial intelligence increasingly influences all of them simultaneously.

Climate modelling relies on AI. Healthcare systems rely on AI. Economic competitiveness increasingly depends upon AI. Military planning increasingly incorporates AI. Public administration increasingly integrates AI into decision-making. International negotiations themselves increasingly rely on AI-assisted analysis.

This changes the nature of diplomacy. Future diplomats may not simply negotiate AI governance. They will negotiate in environments where AI actively supports policy analysis, summarizes negotiations, drafts briefing papers, identifies areas of consensus and proposes alternative negotiation strategies.

AI therefore becomes not only the object of diplomacy. It increasingly becomes part of diplomacy's operating environment.

This raises another governance question that deserves greater attention. As AI systems become more deeply embedded within public institutions, governance can no longer focus exclusively on technological risk. It must also consider institutional judgment.

Every AI system reflects choices made during its development — choices about data, about alignment, about evaluation, about acceptable behaviour, and about priorities. No AI system is entirely neutral.

This does not imply that AI should not be used in diplomacy or government. Quite the opposite. Its potential to improve analysis, increase efficiency and support evidence-informed policymaking is becoming increasingly evident.

However, institutional adoption should never imply institutional dependence. Governments remain responsible for political judgment. Diplomats remain responsible for negotiation. Public institutions remain responsible for accountability.

AI may augment these responsibilities. It cannot replace them.

Perhaps this is where the next chapter of AI governance begins. Not with another declaration of principles. Not with another list of ethical recommendations. But with the careful construction of an international governance architecture capable of connecting diverse institutions, different governance traditions and multiple policy communities.

The future of AI governance may therefore depend less on creating a single global authority than on building mechanisms that allow many institutions to work together coherently.

In many respects, this resembles diplomacy itself. Diplomacy has never required complete agreement. It has always depended upon creating enough common ground for cooperation to remain possible despite differences.

Perhaps AI governance will evolve in much the same way — not as a single system, but as an interoperable architecture built upon dialogue, coordination and shared responsibility. If this trajectory continues, the defining challenge of the coming decade may not be building more capable AI systems. It may be building governance institutions capable of keeping pace with them.

During the United Nations Global Dialogue on AI Governance meeting, one question stayed with me throughout the discussions:

Can the international community truly converge around a shared vision for governing artificial intelligence?

After two days of listening to scientists, heads of state, ministers, diplomats, and representatives of international organizations, one conclusion became increasingly clear. Artificial intelligence is no longer viewed simply as a technological breakthrough. It has become a strategic issue at the intersection of diplomacy, economics, security, sustainable development, human rights, and international cooperation.

The dialogue opened with the presentation of the Independent International Scientific Panel's report, prepared by 40 experts from different disciplines and regions. The report highlighted AI's enormous potential to accelerate progress in healthcare, education, science, and agriculture, while also presenting a sober assessment of the challenges ahead — from cybersecurity and information integrity to child safety, environmental sustainability, and growing global inequalities. Perhaps its strongest message was that the future of AI will depend less on technological capability than on the governance choices we make today.

The governmental sessions that followed offered an equally valuable perspective. While national priorities naturally differed, there was remarkable convergence around several core principles. Trust, human-centred AI, international cooperation, capacity building, interoperability, and responsible governance emerged repeatedly, regardless of region or political system.

At the same time, countries emphasized different aspects of the AI agenda. European governments focused largely on trustworthy AI, human rights, democratic resilience, and regulatory frameworks. China highlighted open-source innovation, AI capacity building, and support for developing countries. Gulf countries emphasized compute infrastructure, energy, and long-term investment, while others stressed institutional capacity, digital sovereignty, and international legal frameworks. Rather than representing competing visions, these perspectives reflected the multidimensional nature of AI governance itself.

One of my strongest observations was the emergence of what could be described as a common diplomatic vocabulary around AI. Concepts such as trust, safety, interoperability, capacity building, international cooperation, and human oversight appeared consistently throughout the discussions. While significant differences remain regarding implementation, the foundations of a shared global conversation are beginning to take shape.

Another notable shift was that AI is no longer being discussed solely through the lens of innovation. Compute capacity, data infrastructure, energy consumption, open-source ecosystems, child protection, cultural diversity, democratic integrity, and environmental sustainability have all become integral parts of the governance conversation. AI governance is increasingly evolving into a multidisciplinary and multi-stakeholder policy domain rather than a purely technical discussion.

After the meeting, I was left with the impression that this dialogue represented not a conclusion, but the beginning of a much larger process. The broad principles are becoming increasingly visible. The real challenge now lies in translating those principles into practical institutions, international standards, and effective mechanisms for cooperation.

Ultimately, the future of AI will not be determined by technology alone. It will be shaped by the collective decisions of governments, international organizations, academia, the private sector, and civil society working together.

Perhaps that is the most important takeaway from Geneva. In the age of artificial intelligence, international cooperation is no longer simply desirable — it is becoming an essential condition for ensuring that AI serves humanity as a whole.

Origin of this Policy Note

This article builds on the ideas presented during my paper at the European Negotiation Conference (ENEA 2026) hosted by Sciences Po Paris, titled "AI, Negotiation and Trust in 2040: Human-Centered AI Diplomacy." Rather than reproducing the conference presentation, this policy note further develops its central argument and introduces an emerging conceptual perspective on the future of AI diplomacy.

Executive Summary

Artificial intelligence is becoming deeply embedded in international affairs. Governments are already using AI to strengthen intelligence analysis, strategic foresight, policy modelling and decision support. As these capabilities mature, AI will inevitably reshape diplomacy itself.

Yet diplomacy has never been solely about information. It has always been about trust.

This policy note argues that the next frontier of AI diplomacy is not the automation of negotiations, but the preservation of human legitimacy within increasingly AI-enabled decision-making environments.

To address this challenge, this article introduces the initial concepts behind the Human-Centred AI Diplomacy (HAD) Framework — a conceptual approach exploring how artificial intelligence can strengthen diplomacy while preserving human judgement, institutional trust and political legitimacy.

A more detailed AI Statecraft Working Paper further developing the HAD Framework is currently in development.

The Wrong Question About Artificial Intelligence

Whenever artificial intelligence is discussed, one question almost always dominates the conversation: how intelligent will AI become?

Governments ask it. Technology companies ask it. Investors ask it.

While this question remains important, I increasingly believe it is not the one that will define the coming decades.

How human will diplomacy remain as artificial intelligence becomes increasingly intelligent?

Artificial intelligence is no longer transforming technology alone. It is reshaping economies. It is redefining geopolitical competition. It is influencing national strategies. And increasingly, it is transforming diplomacy itself.

AI Is Not Just Another Topic of Negotiation

For many years, diplomacy focused on territory, trade, energy, security and resources. Artificial intelligence is often presented as simply another item added to that agenda. I believe this perspective is incomplete.

AI is not merely becoming another subject governments negotiate. It is beginning to transform how negotiations themselves are conducted. Artificial intelligence changes how information is generated, how risks are assessed, how scenarios are simulated, how decisions are supported, and ultimately how trust is established.

This represents a much deeper transformation than simply negotiating AI regulation. It represents a transformation of diplomacy itself.

Intelligence Is Becoming Abundant

For centuries, intelligence represented one of humanity's greatest strategic advantages. Today we are creating systems capable of analysing vast amounts of information within seconds. Knowledge is becoming increasingly abundant.

If intelligence becomes abundant — what becomes scarce? Humanity.

Not because humans possess more information than machines. But because diplomacy has never depended solely upon information. It depends upon judgment, empathy, political responsibility — the ability to understand intentions that remain unspoken, and to create legitimacy among actors whose interests fundamentally differ. These are not soft skills. They are strategic capabilities.

Diplomacy Is Built on Trust, Not Data

Artificial intelligence can optimise negotiations. It can recommend strategies. It can evaluate thousands of policy alternatives. It can forecast risks that no individual analyst could identify alone.

Yet successful diplomacy has rarely depended upon having more information than others. Successful diplomacy depends upon creating sufficient trust for actors to move beyond uncertainty together.

AI can reduce uncertainty in data. It cannot eliminate uncertainty in human intention. And diplomacy ultimately exists to manage human intentions — not datasets.

Perhaps this is the greatest misconception surrounding AI diplomacy today. The objective should not be replacing diplomatic judgement. The objective should be strengthening it.

Towards Human-Centred AI Diplomacy

These reflections have led me toward what I describe as Human-Centred AI Diplomacy. Rather than asking whether AI will replace diplomats, I believe we should ask: how can AI strengthen diplomacy while preserving what makes diplomacy fundamentally human?

Human-Centred AI Diplomacy is not an argument against artificial intelligence, nor an argument for slowing technological progress. Instead, it proposes that AI should enhance diplomacy while human beings remain responsible for legitimacy, accountability, political judgement and trust.

Technology may accelerate diplomacy. Only people can legitimise it.

Introducing the HAD Framework

Building on this perspective, I am currently developing the Human-Centred AI Diplomacy (HAD) Framework as a conceptual model for understanding the evolving relationship between artificial intelligence and diplomacy.

The framework explores how AI can augment diplomatic capability without diminishing human responsibility. Rather than positioning AI and diplomacy as competing forces, it seeks to understand how technological capability, institutional trust and human judgement can evolve together.

This policy note introduces the initial concepts behind the HAD Framework. A more detailed AI Statecraft Working Paper expanding the framework is currently in development.

Looking Toward 2040

By 2040, diplomats may routinely work alongside AI systems capable of strategic forecasting, multilingual negotiation support, geopolitical simulations and real-time policy analysis. The institutions of diplomacy will undoubtedly evolve. But diplomacy itself should not lose its human foundation.

History may remember those who built the most powerful AI systems. I believe it will also remember those who ensured that those systems remained accountable to people, institutions and shared human values.

Artificial intelligence may transform diplomacy. But diplomacy must continue to shape artificial intelligence. Because the future of international cooperation will depend not only on intelligent systems — but on our ability to remain deeply, responsibly and fundamentally human.

What if NATO's future is defined not only by stronger AI systems, but by the trust shared among Allies?

Over the past week, through discussions in Brussels and Paris and my participation in international policy and diplomacy forums, one common theme became increasingly clear: artificial intelligence is no longer merely a domain of technological competition—it is becoming a strategic pillar of security, diplomacy, and international cooperation.

As the NATO Summit convenes in Ankara on 7–8 July, one question continues to stand out in my mind:

What will be the defining strength of an alliance in the age of artificial intelligence?

For decades, security has been measured through military capability, deterrence, and technological superiority. Today, however, that equation is evolving. Artificial intelligence is becoming deeply embedded across defence ecosystems — from decision-support systems and intelligence analysis to cyber defence, logistics, and autonomous capabilities.

Yet this transformation raises a new strategic question.

How can Allies trust one another's AI systems?

High-performing algorithms alone will not be enough. AI systems must also be understandable, transparent, auditable, and subject to meaningful human oversight. Equally important, they must be capable of operating reliably across different national systems and institutional environments.

In other words, technological capability without trust cannot sustain an alliance.

For this reason, I believe NATO's next strategic challenge will not simply be developing more advanced AI capabilities, but strengthening the governance mechanisms that ensure those capabilities are trustworthy, explainable, interoperable, and responsibly deployed across the Alliance.

NATO has already taken significant steps by updating its AI Strategy, adopting Responsible AI Principles, and launching initiatives such as DIANA and the NATO Innovation Fund to accelerate emerging technologies. These developments demonstrate that AI has become an important strategic priority.

However, I believe the next chapter is less about the technology itself and more about how that technology is governed, trusted, and shared among Allies.

I describe this emerging concept as an "AI Trust Architecture for NATO."

By this, I do not mean merely a set of technical standards. Rather, I envision a comprehensive framework encompassing AI governance, data governance, transparency, accountability, human oversight, assurance mechanisms, interoperability, and shared principles for responsible AI deployment across the Alliance.

Ultimately, the defining advantage of future alliances may not lie in possessing the most advanced AI models. It may lie in creating AI systems that Allies trust enough to rely upon collectively.

In that sense, the Ankara Summit represents more than another high-level diplomatic gathering. It offers an opportunity to begin a broader strategic conversation about how trust should be built in the age of artificial intelligence.

Because in the AI era, collective security will depend not only on stronger technologies, but on stronger trust.