Bahar Serçin Halefoğlu

On 13 June 2026, President Recep Tayyip Erdoğan announced the Türkiye Artificial Intelligence Action Plan 2026–2030. With the Presidential Circular published on 18 August, the Plan has now entered a new phase of implementation. For Türkiye, I believe this development is worth considering not simply as the publication of another technology policy document, but as part of a broader strategic transformation in which artificial intelligence is increasingly connected with economic capacity, public administration, technological sovereignty, security and international positioning.

One of the most notable aspects of the Plan is that it does not approach artificial intelligence solely through models or software development. Türkiye's framework is built around five interconnected layers: energy, computing hardware, infrastructure, models and data, and applications. Talent, financing, and regulation and trust are positioned as cross-cutting components supporting the entire structure. This architecture reflects an important reality: competition in artificial intelligence can no longer be reduced to a single technological domain. Developing capable models matters, but so do the energy infrastructure that supports them, accessible computing capacity, high-quality data, skilled human capital, an enabling investment environment and trustworthy governance.

In this context, the Plan's four pillars — Recognise, Utilise, Produce and Govern — are more than a choice of headings. Together, they provide a progression for Türkiye's AI maturity journey. The first stage is about enabling society and institutions to understand artificial intelligence. The second concerns using the technology effectively. The third is about translating adoption into domestic production and economic value. The fourth focuses on managing this capacity in a secure, sustainable and internationally relevant manner. The Plan's 16 priority actions are structured around this progression.

Türkiye's journey in artificial intelligence, of course, did not begin today. The new Action Plan builds on the 2021–2025 National Artificial Intelligence Strategy, the 2024–2025 Action Plan, the work of the Parliamentary Artificial Intelligence Research Commission and the policy work undertaken by the Presidential Science, Technology and Innovation Policies Board. The preparation process itself is also significant. Workshops were conducted with nearly 500 professionals from the public sector, private sector, universities and civil society, while more than 2,200 action proposals were received from citizens. This suggests that the Plan is intended not simply as a centrally developed policy document, but as a broader national roadmap informed by a wide range of stakeholders.

Human capital is an important part of this approach. The Plan aims for a total of five million people to complete AI literacy programmes within the first 24 months, while also targeting the development of 10,000 advanced AI specialists and 100,000 AI application professionals. The significance lies not only in the scale of these numbers. An assessment system integrated with e-Government is envisaged to evaluate citizens' AI competency levels and guide them towards development programmes suited to their needs. Training is also intended to reach a broad range of groups, from students and public-sector employees to blue-collar workers and industry professionals.

This approach moves AI literacy beyond a narrow area of technical expertise and positions it as part of the social and economic infrastructure of the future. In the years ahead, differences between countries may emerge not only from access to technology, but also from the capacity of societies to understand, use and transform it into value.

The treatment of computing capacity as strategic infrastructure is equally important. Türkiye aims to reach 1 GW of installed data-centre capacity by 2030, mobilise USD 10 billion in predominantly private-sector AI and data-centre investment, and create access to up to 10 million GPU-hours annually through the "GPU for Everyone" approach. These ambitions indicate that computing capacity is increasingly being viewed not simply as a technical requirement, but as an issue closely linked to economic resilience and technological sovereignty.

There is an important balance in Türkiye's approach here. The Plan does not frame digital sovereignty as disengagement from the international system. Domestic data centres, public reserve capacity, international cloud partnerships and regional capacity partnerships are instead envisioned as complementary components. Türkiye's participation in initiatives such as the Digital Europe Programme and EuroHPC is also considered within this wider computing ecosystem.

This matters because technological sovereignty and international interoperability do not necessarily represent opposing choices. When designed carefully, they can reinforce one another. A particularly valuable form of strategic capacity may lie in reducing critical dependencies while remaining strongly connected to global technology, research and investment ecosystems.

The public-sector dimension of the Plan is also notable. A "scan–pilot–scale" cycle is intended to be introduced across public institutions, with high-potential use cases systematically identified, rapid pilots initiated and successful solutions scaled within six to twelve months.

This positions the public sector not only as a regulator of artificial intelligence, but potentially also as an early customer and reference market for domestic technological solutions. An AI solution developed in Türkiye and successfully deployed in the public sector may later enter international markets with a stronger reference base. In this sense, public procurement can also become an instrument of technology and industrial policy.

Türkiye's ambition to become not only a user of artificial intelligence but also a developer and exporter becomes particularly visible in the production dimension of the Plan. Through AI Growth Zones, Regional Centres of Excellence, a National AI Research Fund and an AI Growth Fund, the framework envisages support extending from fundamental research to the scaling of technology ventures. At least TRY 10 billion is envisaged for the Research Fund and TRY 15 billion for the AI Growth Fund, alongside a goal of mobilising at least two units of private capital for every unit of public resources.

It is particularly valuable that Türkiye's existing sectoral strengths are connected with its AI ambitions. The Plan foresees the development of sectoral foundation models in areas including health, drug discovery, genomics, disaster management, energy, agriculture and advanced materials. Work on Turkish-language large language models is expected to continue, while Türkiye's defence-industry and manufacturing capabilities are also intended to support progress in physical AI and robotics.

Türkiye's seismic data, agricultural diversity, healthcare infrastructure, manufacturing capacity and experience in defence technologies may provide meaningful advantages in selected vertical domains. The opportunity for Türkiye, therefore, may not lie solely in competing to develop the largest general-purpose model. It may also lie in transforming its own data, industrial expertise and sectoral capabilities into commercially viable and exportable AI solutions.

The Plan's concept of an "export-ready package" is particularly interesting in this respect. By bringing together security testing, legal compliance, performance documentation and environmental impact requirements within a standard framework, the approach is intended not only to support model development but also to help prepare technologies for acceptance in international markets.

Physical AI and robotics deserve particular attention as part of this picture. Global AI competition is not limited to large language models. Autonomous systems, robotics, intelligent manufacturing, edge computing and industrial automation are likely to become increasingly significant areas of competition.

Türkiye may have a particularly relevant foundation in this area. Its experience in autonomous systems within the defence industry, its automotive and machinery manufacturing capabilities, engineering base and broader industrial ecosystem can provide an important platform for physical AI. The Plan therefore includes initiatives ranging from smart-factory test lines and agricultural robots to disaster and search-and-rescue simulation environments, autonomous-driving test tracks and edge-computing validation laboratories.

The proposed transfer of capabilities from defence technologies towards civilian applications is also accompanied by attention to dual-use risks, export controls, data security and human oversight. This combination of innovation and governance is important.

In my view, physical AI may become one of the areas where Türkiye can develop a differentiated international position. The next phase of global AI competition is unlikely to be defined only by the question of who can develop the largest model. Increasingly, another question will matter as well:

Who can integrate artificial intelligence most effectively into the physical world?

For me, one of the most significant dimensions of the Plan is its fourth pillar: Govern.

Here, AI governance is not confined to regulation or ethical principles. The same strategic framework brings together attracting global investment, conducting international AI diplomacy, developing regional capacity, strengthening trustworthy AI assessment and ensuring model security.

This reflects an important evolution in how artificial intelligence can be understood at the national level. Governing AI is no longer only about determining which rules apply to which systems. It is increasingly connected to investment policy, security policy, industrial policy, standards, international cooperation and state capacity.

Within this context, I find the explicit inclusion of AI diplomacy as part of the national strategy particularly significant.

The Plan envisages Türkiye playing an active role in shaping discussions and standards through the OECD, G20, United Nations and regional platforms. More specifically, by the end of 2027, the Plan aims for at least five bilateral AI cooperation framework agreements, contributions in OECD, G20, UN, ISO/IEC, ITU and UNESCO processes through technical views, position papers, joint statements or working-group participation, memoranda of understanding with at least three international organisations, and the initiation of mutual-recognition or conformity cooperation for Turkish AI products and models in at least three target markets.

What is especially valuable here is that the success of AI diplomacy is not intended to be measured only by the number of agreements or declarations. The Plan also refers to technical working groups, joint testing protocols, mutual-recognition mechanisms, joint calls for projects, and frameworks for data and computing sharing.

This takes AI diplomacy beyond symbolic international engagement. The next generation of technology diplomacy is likely to be shaped increasingly through standards, computing partnerships, data governance, mutual-recognition mechanisms, shared models, research networks and investment arrangements. Türkiye's decision to position these elements directly within its national AI strategy therefore represents an important development.

The regional-capacity dimension is equally noteworthy. Cooperation is envisaged with the Organization of Turkic States, the Organisation of Islamic Cooperation, D-8, Gulf countries and Türkiye's wider neighbourhood, including work on shared data-centre investments, computing capacity and data-governance models.

One of the most concrete examples is the planned development of a shared large language model for Turkic languages. The model is expected to progressively include the Oghuz, Kipchak and Karluk language families, make use of shared HPC capacity, and support areas such as public services, education, translation, access to information and cultural heritage.

This provides a useful illustration of what the future of AI diplomacy may look like.

Interoperability is no longer only about agreeing on common principles. It can also involve building technological capacity together.

If such initiatives develop successfully, Türkiye's geographical and diplomatic position may increasingly support a role not only as a provider or adopter of technology, but also as a regional partner in AI infrastructure, models and governance capacity.

The trustworthy-AI dimension completes this broader architecture. The Plan envisages differentiated requirements based on risk levels, algorithmic impact assessments, model cards, security testing and regulatory experimentation environments. Governance is framed not as an obstacle to innovation, but as an enabling structure for scaling innovation more safely.

This distinction is important. Trustworthiness can increasingly become not only an ethical consideration, but also an economic and strategic asset. Companies and investors are likely to consider not only infrastructure costs and talent availability, but also the predictability, security and international interoperability of regulatory environments.

The emphasis on measurable implementation is another important feature of the Plan. A National Artificial Intelligence Board is envisaged as the highest-level body for strategic direction and coordination, while a Programme Office is intended to follow implementation. A National AI Ethics Board is expected to provide ethical guidance and societal assurance, with technical security assessment capabilities strengthened through cooperation between the Cyber Security Presidency and TÜBİTAK Artificial Intelligence Institute.

The National AI Progress and Transparency Portal is also intended to make progress on targets, budgets, timelines and performance indicators publicly visible. Measurement is not limited to infrastructure. Data, talent, adoption, trust and regulation, investment and exports, as well as international and regional impact, are all expected to be monitored through distinct indicators. The first public progress report is planned within twelve months, while an independent evaluation component is expected to be included in annual reporting.

This is where the broader significance of Türkiye's Artificial Intelligence Action Plan becomes particularly visible.

Energy influences computing capacity. Computing enables research and entrepreneurship. Data supports models. Models become applications across public services and industry. Financing allows successful solutions to scale. Governance creates confidence. Diplomacy connects these capabilities with the international system.

These are not isolated policy areas. When considered together, they represent something broader than an AI ecosystem. They begin to form a national AI capability.

This is also where I see the connection with the idea of AI Statecraft.

Over the past few years, global discussions have repeatedly emphasised that AI is geopolitical, that it is transforming economic competition and that it has become a strategic technology. All of this remains true. But the conversation can now move further.

So what comes next? How can a country translate that strategic importance into tangible capacity? How can energy, data, compute, talent, public institutions, industry, financing, governance and diplomacy be connected?

For me, one of the most important aspects of the Türkiye Artificial Intelligence Action Plan is precisely this. Rather than simply reiterating the strategic importance of artificial intelligence, it places these different dimensions within a more integrated policy architecture.

The extent to which every objective translates into results will naturally become clearer through implementation over the coming years. Yet, as Türkiye enters the 2026–2030 period, the decision to approach artificial intelligence not only as a global technology to be adopted, but as a national capacity to be developed, produced, governed responsibly and strengthened through international cooperation is highly significant.

Türkiye's young human capital, manufacturing and defence-industry experience, entrepreneurship ecosystem, digital public infrastructure, regional connections and growing technological capabilities offer a valuable foundation for this transformation. The opportunity ahead is to connect these strengths within one strategic ecosystem rather than allow them to remain separate success stories.

Because the next stage of global competition in artificial intelligence is unlikely to be determined solely by who possesses the most powerful model. It will also be shaped by another question:

Who can bring together the strongest talent, infrastructure, industrial capacity, trusted institutions and international partnerships?

This is where an important part of the next strategic competition is likely to take place. The Türkiye Artificial Intelligence Action Plan 2026–2030 represents an important and promising step towards strengthening Türkiye's position as a more capable, productive and internationally engaged AI actor.

This is where the transition from vision to capability begins.

This article reflects the author's personal assessment and is intended to contribute to the broader discussion on artificial intelligence, governance, diplomacy and strategic technology.

For the past two years, one sentence has dominated almost every international discussion on artificial intelligence:

AI is geopolitical.

It is repeated in policy papers, diplomatic forums, international summits and government strategies. It reflects a reality that is difficult to dispute. Artificial intelligence is no longer simply a technological innovation. It has become a strategic capability influencing economic competitiveness, national security, industrial policy, diplomacy and the global balance of power.

I fully agree with this assessment. Yet I increasingly believe that it is no longer enough.

Recognising AI as a geopolitical issue explains why artificial intelligence matters. It does not explain what should happen next. In many ways, "AI is geopolitical" has become the starting point of the conversation rather than its conclusion.

Over the past months, through international dialogues, discussions with policymakers, diplomats, scientists and governance experts, I have noticed an important shift. Surprisingly, very few conversations are still about convincing people that AI is strategically important. That debate appears to have matured. Instead, the discussions are becoming significantly more operational.

The questions being asked today are fundamentally different from those we heard only a few years ago.

Rather than debating whether AI should be transparent, discussions increasingly focus on how transparency can be measured. Rather than asking whether AI systems should be accountable, policymakers ask which institutions should be responsible for ensuring accountability. Rather than discussing whether safety matters, attention is turning towards evaluation methodologies, independent testing, auditing mechanisms and continuous oversight. Instead of debating the importance of international cooperation, conversations increasingly revolve around capacity building, governance implementation, scientific assessment, institutional coordination and common technical standards.

This may appear to be a subtle evolution.

I believe it is anything but subtle. It represents the transition of AI governance from a normative conversation into an institutional one.

For several years, the international community invested enormous effort in establishing a common governance vocabulary. Trust. Transparency. Accountability. Human oversight. Fairness. Safety. These principles remain indispensable. They created a shared language that enabled governments, international organisations, academia and industry to begin discussing AI within a common ethical and policy framework.

However, principles alone do not govern technologies. Institutions do. Processes do. Standards do. Scientific evidence does. Implementation does.

Perhaps this is the most important transition currently taking place in global AI governance. The conversation is gradually moving away from defining what responsible AI should look like and towards designing the institutional capacity required to make responsible AI possible.

This distinction matters. Writing principles is only the first step. Turning those principles into measurable, enforceable and internationally coordinated governance mechanisms is an entirely different challenge.

This is where I believe the next phase of AI governance begins.

Increasingly, governance discussions are no longer centred on regulation alone. They encompass scientific panels capable of informing policymaking through evidence, independent evaluation frameworks that assess increasingly capable AI systems, technical standards that enable interoperability across markets and jurisdictions, capacity-building initiatives designed to ensure that developing countries are not left behind, public procurement frameworks that determine how governments themselves deploy AI responsibly, institutional arrangements capable of coordinating public authorities, private companies, researchers and civil society, and diplomatic processes that seek common approaches despite differing political systems and regulatory traditions.

Taken individually, these initiatives may appear unrelated. Collectively, they reveal something much larger.

AI governance is becoming an institutional ecosystem rather than a collection of regulations.

This evolution is hardly surprising. Technologies with systemic impact inevitably generate systemic governance requirements. Artificial intelligence is now influencing healthcare, education, finance, public administration, defence, critical infrastructure, scientific research and democratic processes simultaneously.

Consequently, governance can no longer be understood as a purely technological discussion. Nor can it be reduced to legal regulation alone. AI governance increasingly requires scientific expertise, diplomatic negotiation, institutional design, technical standardisation, economic strategy and international coordination operating together.

In many respects, governance capacity itself is becoming a strategic capability.

This observation also changes how we think about international competition. Much attention continues to focus on frontier models, compute capacity, semiconductor supply chains and investment volumes. These factors undoubtedly matter.

Yet another competition is quietly emerging alongside technological competition.

The competition to build trusted governance capacity.

Countries capable of developing credible institutions, independent oversight, effective implementation mechanisms and internationally trusted governance frameworks may ultimately exercise influence comparable to those leading AI development itself.

Technological leadership and governance leadership are becoming increasingly interconnected. One without the other may prove insufficient.

This is particularly important because AI governance is no longer confined to questions traditionally associated with technology policy.

Data governance. Cybersecurity. Energy infrastructure. Environmental sustainability. Child safety. Public sector transformation. Scientific integrity. International development. Digital inclusion. Defence.

These are no longer separate policy discussions. They are becoming interconnected dimensions of a single governance agenda.

Perhaps this explains why AI diplomacy is also evolving. Its role is no longer limited to negotiating broad ethical declarations or political commitments. Increasingly, AI diplomacy is becoming the practice of building common governance capacity across different political, regulatory and institutional systems.

Its objective is not necessarily to create identical governance models. Rather, it is to enable countries with different legal traditions, economic priorities and strategic interests to cooperate on issues that none of them can effectively manage alone.

That may ultimately become one of the defining diplomatic challenges of the coming decade.

For this reason, I believe we should begin asking different questions.

Instead of asking whether AI is geopolitical, perhaps we should ask whether our governance institutions are evolving at the same pace as the technology itself.

Instead of debating whether AI requires governance, perhaps we should ask what kinds of institutions, scientific mechanisms and diplomatic frameworks are capable of governing technologies that evolve faster than traditional policymaking cycles.

And instead of assuming that stronger AI alone will determine global leadership, perhaps we should recognise that governance capacity may become an equally important source of strategic influence.

Artificial intelligence has already changed geopolitics.

The next question is whether global governance can evolve quickly enough to manage that new reality. In my view, this is where the next chapter of AI Statecraft truly begins.

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.