AI has already redrawn the map of global conflict. The real challenge is to build a governance architecture suited to the speed and complexity of this transformation. Join us on Telegram , Twitter , and VK . Contact us: info@strategic-culture.su Dual-use as a structural condition When, in 2011, mathematician and entrepreneur Eric Schmidt—then at the helm of Google—described artificial intelligence as “the most transformative technology humanity has ever developed,” few could have imagined how quickly that prediction would translate into military doctrine. Today, just over a decade later, AI is not merely a tool that enhances the effectiveness of armies: it has become the very mechanism through which military power is generated, organized, and projected in the contemporary world.
We have entered a new era of conflict, in which the boundaries between war and peace, between civilian technology and military weaponry, between kinetic operations and digital influence campaigns, have become as thin as an algorithm. And with each passing day, that line becomes harder to draw.
The first element that distinguishes AI from any previous technological innovation—from gunpowder to nuclear power, to drones—is its inherently dual-use nature. Unlike traditional weapons systems, designed from the outset for military purposes, artificial intelligence systems almost always originate within commercial and civilian ecosystems: university labs, tech startups, major digital platforms.
A language model developed to assist a bank’s customers can, with relatively few modifications, be used to analyze enemy communications or generate disinformation on an industrial scale. A computer vision system designed to drive autonomous cars can be repurposed for military target recognition. A logistics optimization algorithm built to manage commercial supply chains can plan troop movements and calculate resupply routes.
This dynamic produces an unprecedented structural condition: the distinction between the civilian and military spheres is becoming increasingly blurred and unstable. Private investment in AI research and development now far exceeds government budgets dedicated to the same sector, meaning that major technological innovations in the military field come—directly or indirectly—from private companies operating in the global market.
This leads to a paradoxical geopolitical consequence: civilian AI technologies can no longer be considered politically neutral tools. Once adapted and deployed in conflict contexts, these systems become integrated into military logic and progressively acquire strategic functions that their creators often neither anticipated nor intended. NATO organizes: from Ukraine to DIANA At the institutional level, the West has grasped the stakes relatively quickly. In the context of the war in Ukraine—the first major conventional conflict of the 21st century in which AI has played a leading operational role—the United Kingdom has actively supported the creation of a Center of Excellence for Artificial Intelligence at the Ukrainian Ministry of Defense. An initiative that is not only a show of support for Kyiv but an open-air laboratory for testing the doctrines, capabilities, and vulnerabilities of AI systems under real-war conditions.
The Atlantic Alliance as a whole has adopted a NATO Strategy on Artificial Intelligence, a document that explicitly recognizes AI as a fundamental enabling technology for contemporary military operations. At the same time, NATO has launched the Defence Innovation Accelerator for the North Atlantic, known by the acronym DIANA: a transnational network designed to connect startups, research centers, and defense industry actors in the shared development of dual-use technologies, with artificial intelligence systems at the top of the agenda.
DIANA represents an attempt to institutionalize the permeability between the civilian and military worlds that AI has made inevitable, transforming it from a vulnerability into a strategic advantage. The idea is that Western powers can maintain a technological edge over their adversaries by accelerating innovation cycles and reducing the time between a scientific discovery and its operational application in the field.
The integration of AI into command and control systems, intelligence analysis, and target identification procedures is already an operational reality. Military decision-making processes are accelerating at a pace that challenges human oversight capabilities: algorithms analyze enormous volumes of satellite, electronic, and signal data in real time, providing operational recommendations within increasingly compressed time frames. The question of how much weight human judgment should carry in this chain—and at what point in the chain it should intervene—is today one of the most pressing issues in the international security debate. Venezuela and Iran: the first laboratories of geopolitical AI While the institutional dimension of military AI is being developed in NATO corridors and Western research centers, it is in real-world crisis theaters that new doctrines are being tested. Two cases, in particular, have caught analysts’ attention as the first documented examples of large-scale strategic operations based on artificial intelligence: Venezuela and Iran.
During the Venezuelan crisis of 2026, AI technologies were reportedly employed by the United States in the context of non-kinetic operations—that is, operations that do not involve the direct use of military force—which included cyberattacks, social media influence campaigns, and disruptions of electromagnetic communications.
What makes the Venezuelan case particularly significant is not so much the nature of the individual operations—which had already been tested in various forms in previous contexts—as their integration into a unified, data-driven framework. Intelligence gathering, surveillance activities, financial tracking operations, and digital influence campaigns are orchestrated in a coordinated manner, fueled by a continuous stream of data analyzed in real time by artificial intelligence systems.
The most revealing detail concerns the nature of the tools employed. Advanced generative AI systems—including, according to some sources, models such as Claude developed by Anthropic—are said to have been integrated into operational environments via platforms built by Palantir, the American company specializing in data analysis for the defense and intelligence sectors. This indicates something structurally new: commercial foundational models, developed and distributed for civilian use, are being progressively incorporated into military intelligence and targeting architectures.
Venezuela thus offers the first concrete illustration of what we might call the “geopolitical toolkit of the algorithmic era”: an integrated instrument that combines intelligence capabilities, sanctions enforcement mechanisms, cyber operations, and information manipulation within a unified, persistent, and scalable operational framework.
While the Venezuelan case illustrates a model of low-intensity conflict mediated by AI, the Iranian context pushes this logic toward more intense and direct forms. The confrontation between Iran, Israel, and the United States has progressively extended into the digital sphere, with operations targeting not only critical infrastructure but also the digital platforms used daily by the civilian population.
Available reports describe cyber operations conducted with the aim of compromising mobile applications widely used by Iranian citizens, in order to disseminate coordinated messages intended to shape public perception and influence behavior. This represents a significant qualitative leap compared to traditional disinformation operations: AI allows for the mass-scale personalization of messages, tailoring them to individual user profiles, their digital behavior patterns, and their social networks.
Together, the Venezuelan and Iranian cases represent more than mere episodes of contemporary geopolitics. They are prototypes of a new form of conflict in which artificial intelligence does not serve as a tool to support operations designed according to traditional logic, but becomes the structuring component of the entire operational cycle: from data collection to planning, from execution to the assessment of effects. Nuclear proliferation is reborn in digital form Alongside operational applications, the integration of AI into the security domain raises an issue that Western governments are beginning to address with growing urgency: the risk that advanced AI systems could facilitate the development of weapons of mass destruction, particularly chemical, biological, or nuclear ones.
These are not science-fiction scenarios. Advanced AI systems, trained on massive corpora of scientific literature, are capable of suggesting synthetic pathways for dangerous chemical compounds, simulating the behavior of biological agents, and optimizing the design of explosive devices. Existing safeguards—the so-called “guardrails” built into the models—are designed to limit these outcomes, but the developers themselves acknowledge that they cannot offer absolute guarantees.
Sophisticated prompt engineering techniques—that is, manipulating the way requests are formulated to the system—can in some cases circumvent the intended protections. The potential leakage of classified information through AI systems integrated into sensitive operational environments adds an additional layer of risk.
The consequence is that AI security can no longer be treated exclusively as a technical issue—as an engineering problem to be solved with better code and more effective filters. It is increasingly a matter of non-proliferation and international security governance, requiring tools analogous to those developed in the post-war period for nuclear arms control: international treaties, verification mechanisms, inspection regimes, and shared norms on state responsibility. The governance that is missing Yet, precisely in the realm of international governance, the framework appears deeply lacking today. There are no binding international conventions that prohibit or substantially limit the military application of artificial intelligence. No treaty analogous to the Nuclear Non-Proliferation Treaty, no equivalent of the Chemical Weapons Convention, no binding norm comparable to the Geneva Conventions for autonomous weapons.
What exists, so far, is a set of declaratory principles, voluntary guidelines, and multilateral forums that, while producing important documents, are unable to bind the behavior of states. The Responsible AI in the Military Domain summit—known as REAIM—reiterated, in its most recent edition, that Western-led initiatives continue to prioritize voluntary principles over binding restrictions. An approach that reflects, at least in part, the reluctance of the most technologically advanced powers to give up competitive advantages they struggle to build and fear seeing eroded by less scrupulous competitors.
The situation is exacerbated by a structural paradox: the countries that most strongly advocate for the need for rules are often the very same ones that, in practice, show the least willingness to accept binding restrictions on their own military AI programs. The gap between the rhetoric of responsibility and the reality of investments and operational doctrines is, in this field, abysmal. BRICS and the Global Majority: an alternative path The governance of military AI is not confined solely to the Western sphere. Within the framework of the BRICS and among the countries that identify with the so-called “Global Majority”—a term designating the group of countries in the Global South that refuse to automatically align with the positions of Washington or Brussels—an alternative trajectory is taking shape, shaped by profoundly different concerns and priorities.
At the heart of this trajectory lies the issue of technological sovereignty. Countries not at the forefront of AI development—and they constitute the vast majority—fear finding themselves in a state of structural dependence on systems developed in the United States, China, or Europe: systems that could incorporate biases, backdoors, or surveillance mechanisms alien to their national interests. The issue of unequal access to advanced AI technologies is perceived, in this context, as a new form of technological imperialism, capable of crystallizing and amplifying existing global power hierarchies.
At the same time, many of these countries express resistance to what they describe as an attempt by the dominant powers in the sector to monopolize international AI standards. If the United States and its allies are the ones writing the rules of the game—deciding what constitutes “responsible AI” and what does not, which military applications are acceptable and which must be banned—the risk is that those rules will reflect the interests and values of those who wrote them, rather than a genuinely global consensus.
The result is a recurring tension in international forums dedicated to AI governance: on the one hand, various states emphasize the need for moderation and collective responsibility in the military use of artificial intelligence; on the other, concrete proposals aimed at limiting or banning certain applications fail to garner sufficient consensus, because the leading powers—both Western and non-Western—systematically find political or strategic reasons not to adhere. The Algorithmic Arms Race Against this backdrop of fragile governance and growing geopolitical tensions, the development of military AI capabilities is proceeding among all major global actors. The United States, China, and Russia, as well as emerging regional powers such as India, Israel, Turkey, and South Korea, are investing heavily in autonomous systems, in artificial intelligence applied to intelligence and electronic warfare, and in data analysis platforms for operational use.
The technological competition between Washington and Beijing is particularly intense and is playing out on multiple fronts simultaneously: from research on advanced chips and semiconductors—considered the physical infrastructure of AI—to large language models, from image recognition systems to autonomous command-and-control platforms. Both superpowers have realized that the advantage in military AI is not merely about the ability to fight future wars more effectively: it concerns deterrence, strategic credibility, leverage in diplomatic negotiations, and, ultimately, relative standing in the international order.
This competitive dynamic generates a self-perpetuating arms race: each actor perceives the opponent’s investment as a threat requiring a response, and each response in turn becomes the justification for further investment by the other. A mechanism already familiar from the history of nuclear weapons and the Cold War, but one that in the AI era unfolds at much higher speeds and involves a much broader range of actors.
The risk in this scenario is not necessarily that of a large-scale conventional war. Rather, it is that of a gradual and creeping destabilization: low-intensity conflicts mediated by AI, crises accelerated by algorithmic decision-making processes, and incidents caused by autonomous systems that no one had explicitly authorized to act. Military history is full of wars that began by mistake, misunderstanding, or unintentional escalation. AI, with its ability to compress decision-making times and reduce human presence in the chain of command, could multiply this risk exponentially. So, where are we headed? Artificial intelligence has already redrawn the map of global conflict. This is not a process that is yet to take place: it is an ongoing process, documented in real operational theaters, institutionalized in the strategies of major alliances, and incorporated into the defense budgets of the world’s leading powers.
The real challenge for governments, international institutions, and civil society is to build a governance architecture suited to the speed and complexity of this transformation. An architecture that can balance states’ security needs with the protection of fundamental rights, that ensures equitable access to AI technologies without turning them into tools of domination, and that establishes binding rules on the military use of AI without relinquishing human oversight in decision-making processes concerning life and death.
It is an extraordinarily difficult political task, requiring a commitment to international cooperation that runs counter to the dynamics of fragmentation and rivalry that dominate today. But it is also an urgent task—perhaps the most urgent on the global diplomatic agenda. Because if technology outpaces politics, the consequences could be irreversible.