The new print volume of Grand Continent The enemy that points at us is now available. Edited by Giuliano da Empoli, you can find it at Gallimard or by clicking here — to receive it and support an independent editorial team, consider subscribing to the magazine
On September 1, 2017, as part of the “Knowledge Day,” Vladimir Putin declared: “Artificial intelligence is the future, not only for Russia, but for humanity as a whole. It generates both colossal opportunities and threats whose scope is still hard to grasp. The one who holds a dominant position in this sphere will, at the same time, become the master of this world” .
At first glance, there would therefore be nothing surprising about Andreï Belousov, the Russian Minister of Defense, announcing, in an enthusiastic tone on June 23, that the Russian Armed Forces are increasingly integrating elements of artificial intelligence, robotic complexes, and integrated command systems for use in combat . In a parallel vein, pro-Kremlin journals close to the Ministry of Foreign Affairs publish analytic pieces addressed to a specialized public, laying out how foreign states at the forefront of this technology are progressively institutionalizing AI as a central component of armed conflicts. Russian military and political circles closely monitor the evolutions of the Gospel system, used by Israel to analyze surveillance data in search of people, equipment, and buildings of the opposing camp before relaying target recommendations to bomb, or the Maven Smart System by Palantir, which provides satellite image analyses and decision-support, notably in the context of the Russo-Ukrainian war.
Where does Russia actually stand, and what are its concrete ambitions? It is all the more delicate to navigate these questions because many analyses are primarily driven by political aims, whether it be opposition media eager to minimize Russia’s technical means to highlight the state’s incapacity, or American media waving the red flag of the “digital front” opened by Russia to justify new investments by the U.S. Defense Department.
In practice, the Russian position appears ambivalent. On the one hand, there is a fundamental tension between two options: attaining real supremacy in the field of artificial intelligence; or, failing that, advocating for international restrictions on the use of this technology in order to preserve at least a relative dominance. On the other hand, technological dependence and the relatively weak scientific investment in Russia doom it to fall behind its Western rivals in this domain, while still granting it, nonetheless, a heightened capacity to cause disruption through misinformation and population control.
The Russian Military AI
Artificial intelligence lies at the heart of debates on the evolution of contemporary military tactics and strategy, given the unprecedented possibilities it offers in searching for and processing information (especially when the data is heterogeneous, contradictory, uncertain, and voluminous), in the automated production and transmission of executive commands, and in decision-support, all within extremely tight timeframes. If the invasion of Ukraine accelerated Russia’s adoption of AI elements, the country would have had little choice but to adapt to this new reality. Yet, results do not seem to live up to the hopes of its strategists.
On the Russian side, the main field of AI integration into military operations concerns autonomous and semi-autonomous combat or support means. Russia possesses the heavy combat drone S-70 Okhotnik, designed as a “loyal wingman” for the Su-57 fighter jet, capable of real-time copilot functions. Ongoing operations in Ukraine have also led Russia to raise the autonomy of attack drones: on July 3, an adviser to Ukraine’s defense minister announced that Russian Armed Forces had massively deployed in the Zaporijia region new models of the Molniya drone, equipped only with an onboard computer and a camera that allow it to operate without human piloting—and thus without telecommunications that could be jammed by contemporary methods . The coming weeks will tell whether these models will be a sufficient match for the Hornet drones used by Ukraine, which also rely on artificial intelligence and cause substantial damage to Russian oil installations. By contrast, it seems that Russia’s older models—semi-competent at best—of autonomous ground and naval drones (the Uran systems used in Syria or the Poseidon submarine) do not incorporate AI elements. In short, Russia’s use of this technology remains far behind Ukraine’s, which has taken this turn earlier and more decisively, with Western partners backing it.
One open question concerns Russia’s ability to profit from a new digital decision-support tool based on AI. The “Svod” system was presented in January 2026 by the Russian Defense Ministry as a tool to collect and fuse several intelligence sources (satellite data, aerial photographs, intelligence reports, and open-source data), analyze them, model possible developments of the tactical situation, and propose the most appropriate options to commanders . If its deployment on the front line has reportedly been ongoing since April 2026, it is difficult to assess its effectiveness, at a time when Russia, for the second time that year, has just lost more territory in a month than it had gained .
The uses of AI by Russia remain far behind those of Ukraine, which has taken this turn in a much earlier and more determined way.
Guillaume Lancereau
These considerations should not lead one to think that Russia is uninterested in artificial intelligence. On the contrary, for some analysts, Russia would be, alongside China, one of the major military powers most determined to emphasize “real autonomy of AI” in the coming years, defined as “a set of AI technologies embedded on board a missile, drone, or robot that allow the system to carry out a mission on its own, without human instructions or dependence on external data sources, in configurations where this autonomy provides a decisive advantage” . Russia and China would thus be the most resolute actors to have, in the near horizon, fully autonomous machines designed for environments of intense electronic warfare, whereas Western countries would remain cautious about such potential uses. It should be noted, however, that this reading, indeed proposed by an American reference media, largely seeks to present Russia as a countermodel and a threat to the West, without providing technical elements to support the analysis.
In reality, Russian military AI continues to face a large number of obstacles, beginning with its total dependence on Western components. According to Ukrainian intelligence data, 5,350 electronic components recovered from various Russian armaments and drone platforms show that the vast majority of components directly related to AI functionalities originate from the West. In the three categories involved (sensors, processors, and memory), American-based companies account for a crushing share, up to 69% of storage components. The supply chain for sensors is more geographically diverse, but the United States still leads with about 38% of components, followed by China (about 16%), Japan, and the Netherlands. Switzerland also represents a significant source, especially for processors . The continuation of these imports through third countries signals the permeability of successive “sanctions trains,” but these restrictions—especially concerning semiconductors and microprocessors—still pose a real brake on Russian technological progress in this area.
What about a Chinese substitute? China already supplies about 16% of the sensors identified, and its share is growing as workaround circuits reorganize. But this substitution faces two limits. Technically, China’s most advanced processors still rely on Western lithography equipment and are primarily allocated to Chinese needs. Politically, Beijing could adjust its deliveries to avoid secondary American sanctions and has no strategic interest in equipping its neighbor with genuine technological autonomy: for China, Russia’s dependence is a lever, not a problem to solve.
Several observers have also pointed to certain weaknesses and rigidities of the Russian technical, legal, scientific, and institutional environment. Russian defense industry companies, starting with Rostec and Morinformsystem-Agat, suffer from bureaucratic regulations and mutual distrust between the political power and military leaders. They also face a structural shortage of qualified personnel, just like the central research institutes of the Russian Defense Ministry. In 2021, the first of these institutes offered a research position in applied mathematics and automated command systems with a salary between $400 and $540. The unattractiveness of positions reserved for skilled workers is one of the structural difficulties facing the authorities. Vladimir Putin himself pointed this out in his September 5, 2025 speech at the Eastern Economic Forum, noting that building a Russian economy capable of looking to the future depended on the state’s ability to provide decent remuneration to these professional categories. Projections reveal a clear gap: while those working in the Russian AI sector currently number in tens of thousands, the national strategy calls for needs reaching up to one million operators, technicians, engineers, and programmers by 2030, while planning to graduate only 15,000 people per year.
The state of the civilian ecosystem, the natural nursery of military applications, confirms this diagnosis. Russia does have national language models (Sber’s GigaChat, YandexGPT), but these lag several generations behind their American and Chinese counterparts due to lack of access to last-generation graphical processors, which sanctions make costly and uncertain to import. The computing capacity installed on Russian soil represents a marginal fraction of that in the United States or China, and no credible plan for national production of advanced semiconductors exists in the foreseeable future. Dual-use can work only in one direction: a civil sector that is anemic cannot feed defense.
Finally, corruption and institutional opacity remain a perennial brake on Russian development, including in the field of AI technologies. This situation is one of the main reasons why Russia, which ranks among the top countries in the world for public funding of research per inhabitant, remains at sixty on the Global Innovation Index, behind Armenia, Mexico, and Morocco. The designation of beneficiaries of the 13.6 billion rubles allocated to AI from 2021 to 2026 was done with notable discretion within a government-listed framework. Large Russian companies regularly request hundreds of millions of rubles in funding without being able to justify its use. The structures intended to represent these actors often appear as empty shells whose sole function is to monopolize the market and the public money tied to it .
Admittedly, Russia’s institutional ecosystem is undergoing a transformation. Training centers, specialized military academies, and AI development agencies are under study or already in place. The government is integrating training related to autonomous systems and AI into schools, universities, technical institutes, and other professional tracks. Yet beyond five- or ten-year forecasts, Russia does not currently appear capable of fulfilling its own expectations.
AI Seen From the Kremlin
Experts and Russian officials are aware that artificial intelligence could soon constitute a real revolution in the conduct of war. For this very reason, admitting that Russia occupies only a secondary position in this field would amount to acknowledging that the Russian Armed Forces are about to fall irretrievably behind. Thus, a realistic look at Russia’s concrete capabilities allows pro-Kremlin actors only two sorts of discourse: either minimize the enemy’s real AI effectiveness, or present it as a threat to global peace and security in order to urge international institutions to regulate it.
The first option is evidently the one favored by the ministry of foreign affairs’ journal in February 2026, which gave voice to a certain Aleksei Leonkov, presented as a “military expert.” According to him, the Ukrainian forces’ 2023 counteroffensive failed due to excessive reliance on AI: “They drafted six different scenarios. All of them were winning. Each guaranteed 100% that Ukrainian forces would reach the shores of the Sea of Azov and ‘split our groups in Zaporijia’s axis in two.’ Everything was perfect. Then we watched the three successive waves of this offensive, to put it simply, crash on the same points and suffer a crushing defeat. In other words, our General Staff triumphed over AI. Human intelligence surpassed AI, and by a wide margin. AI simply failed to model correctly the actions of our armed forces or to account for the factor of ‘military cunning’” .
However, no fact can be adduced to support this thesis. Leonkov’s claims are unverifiable, they are not echoed by any other source, and they reduce to a single—quite secondary—factor an overall failure that actually stems from a plurality of causes. If one had to pick only one, it would be the excessive caution and the delays recorded by Western countries in delivering the military equipment Ukraine needs.
A realistic view of Russia’s concrete capabilities leaves pro-Kremlin actors with only two kinds of discourse: minimize the enemy’s real AI effectiveness or frame it as a threat to world peace and security.
Guillaume Lancereau
That argument, difficult to sustain, remains rather marginal in pro-Kremlin publications, which more often advocate a critical approach to how Western states use AI. On January 19, a piece by two Russian professors from Moscow’s State Institute of International Relations, the National Institute for Security Studies, and the Diplomatic Academy outlined the supposedly contradictory logics animating the Western and the “Eastern” conceptions of AI. On the American side, the authors highlighted a series of measures implemented by President Donald Trump between late 2025 and early 2026, concluding that the West had all converted to an “AI-First” logic—i.e., a system organized around algorithms, in which artificial intelligence would be the fundamental backbone of all military operations, from intelligence to logistics. They claimed that the Eastern approach—i.e., China’s—represented the exact opposite: an “open” AI, not privatized, based on regulation, cooperation, and decentralization, counter to hegemonic, competitive, and monopoly-driven logics by companies like NVIDIA and OpenAI. More broadly, they argued that this approach would be favored by the BRICS, who remain sensitive to international equity, whereas the United States would be guided by its own international dominance. The publication’s evident bias becomes apparent when one considers any other field where Russia holds a stronger position—one would anticipate the day when Russian authorities promote an open, decentralized, and equitable approach to the missiles and strategic armaments at their disposal.
The same authors also accuse the United States of withdrawing unilaterally from international initiatives aimed at framing AI uses, particularly military, through a series of United Nations resolutions in 2025. Echoing this publication, two researchers from the Institute of World Economy and International Relations of the Russian Academy of Sciences published in November a synthesis article contrasting the American logic—carefree about any external or internal AI controls—with the doctrine outlined in 2023 by Russia at the Geneva-based UN Group of Government Experts under the Convention on Certain Conventional Weapons. The working paper from the Russian Federation insisted on ensuring the use of AI in a manner compatible with international law, including international humanitarian law. The document concluded with the statement: “By implementing these conceptual guidelines, the Armed Forces of the Russian Federation will commit to using AI-integrated technologies in the interests of national defense, in strict compliance with the norms of international law.” The authors of this synthesis even contemplated raising at the international level the question of treating autonomous and semi-autonomous military systems endowed with AI as weapons of mass destruction, in order to slow the arms race fueled by the great powers.
Control and Disinformation
Because of these technical limits and political ambiguities, some commentators from the Russian opposition tend to conclude that all announcements by the Russian state on AI are mere powder in the wind. Thus, in the pages of Novaya Gazeta in March: “We must state without the slightest hesitation that all government initiatives in this domain belong to pure science fiction. It’s all just noise: a backdrop of flashy declarations, mountains of paperwork, and interminable bureaucratic processes that resemble a simulation of activity more than a real effort.” .
Yet there are two domains in which the Russian authorities’ capacity to mobilize AI gives them real capacity to cause harm: disinformation and the surveillance of their own population.
In terms of disinformation, Russia possesses a solid offensive architecture that could leverage AI integration to its fullest. A researcher from the Belfer Center (Harvard University) noted that models such as Claude Mythos, which are designed to identify and exploit software vulnerabilities, could usher in a new era of cyberwarfare that Russia could approach from a favorable position. It has long maintained semi-autonomous criminal groups protected by state support, along with a proven doctrine of information warfare, all in a context of weak regulatory constraints. Russia would not even need to develop its own tools in-house: it could simply adopt Open Source models as soon as they are published to conduct a range of operations—from cyberattacks to physical sabotage to large-scale disinformation campaigns .
From the state’s perspective, all these operations fall within a strictly defensive logic. The ministry of foreign affairs’ journal even stated in 2023 that the real danger posed by artificial intelligence came from disinformation and manipulation practices by Western countries, whose machinations would compel BRICS to establish a new international cybersecurity cooperation platform . For this same reason, the deputy minister for Digital Development, Aleksandr Choytov, recently announced that AI itself should be subject to “censorship” in Russia, with bans on certain models and certain queries—a proposition clearly unconstitutional, given that the Russian Constitution forbids any form of official censorship in the Russian Federation .
There are, however, two domains in which the ability of Russian authorities to mobilize AI gives them real capacity to cause harm: disinformation and the surveillance of their own population.
Guillaume Lancereau
Here as elsewhere, the official and semi-official Russian discourse amounts to a straightforward inversion of reality. For years now, Russian services have used AI to produce “deepfakes” as part of their information operations. In 2022, videos attributed to Russia depicted President Volodymyr Zelensky announcing Ukraine’s surrender or Kyiv mayor Vitali Klitschko urging European counterparts to repatriate war refugees. Similar methods were systematically deployed by Russia in its bid to reassert influence over Armenia on the eve of elections that ultimately confirmed the country’s pro-European orientation. Likewise, AI fuels the myriad of Russian chatbots designed to generate favorable comments about Russia on social networks.
The latest large-scale operations targeting Armenia, but also Hungary and Moldova, demonstrated both Russia’s interference capabilities and the relative immunity of much of civil society to these manipulations. Yet one must agree on what “failure” means here. As Ian Garner suggests regarding what he calls “slopaganda”—the IA-generated visual slush “produced at scale and disseminated on social networks for political purposes”—the effectiveness of these contents is not measured by their ability to persuade: their main and scariest effect is not simply to flood us with a false reality, but, by saturating us, to create an incessant impression that everything could be false. In this light, Russian campaigns did not necessarily fail because they failed to sway opinion: they merely sow doubt and render any image or statement suspect, including those from pro-European actors, an operation made profitable even at a marginal yield because of its low production cost.
Conversely, the Russian population itself risks bearing more direct consequences from AI embedded in surveillance, censorship, and political control tools. At present, Roskomnadzor already uses AI extensively to monitor and block Russian Internet content, as revealed by a recent data leak caused by Belarusian hackers . Roskomnadzor has been developing for years systems dedicated to several tasks: continuous analysis of social networks and media to detect signals of sensitive topics (mobilization and military losses, protests, LGBT content, criticism of the Russian power); detection of prohibited information in online texts; image and video recognition to identify calls to protest, insulting montages against the president, or symbols banned by the state.
Another particularly sensitive domain is publishing, targeted for months by searches and criminal prosecutions for publication—even retrospectively—of titles deemed contrary to “traditional values.” It is also under the pretext of preserving readership that Roskomnadzor intends to institute a prior censorship system for books: effectively, to “offer” publishers the option to submit their manuscripts to AI to determine their conformity with the new legislation on “drug propaganda” in force in Russia since September 2025. One can easily imagine the market-censoring capabilities this system would offer if it were not limited to controlling narcotics promotion but extended to the ideological validity of works .
In the near future, AI could allow Roskomnadzor to counter one of the most popular strategies for bypassing blockages: VPNs. Russian authorities are investing in a new AI-based algorithm capable of detecting blocked sites’ mirrors by analyzing their content rather than just IP addresses, while detecting encrypted traffic passing through VPNs and slowing access to resources deemed a state threat .
Finally, some analysts foresee the next step as adopting AI-based systems to further clamp down on Russia’s political campaigns and elections—beginning with the autumn 2026 State Duma elections . While the concrete forms these devices might take remain unclear, these forecasts confirm the trend outlined in the preceding sections. AI would not create ex nihilo new spheres of censorship, disinformation, surveillance, or electoral manipulation; it would magnify and extend the reach of practices the Russian state is already employing.
Artificial intelligence represents both a promise and a threat for Vladimir Putin’s Russia: the promise of more capable military means, cyberwarfare tools, and surveillance techniques; the threat of an irreparable lag in the military domain and of increasingly effective retaliation strategies against foreign disinformation and domestic population control attempts.
Questions about AI thus extend beyond a bundle of objective technical capabilities; they touch on how different actors perceive their own technical capacities and those of their adversaries. On the strictly military plane, Jacquelyn Schneider, director of the Hoover Wargaming and Crisis Simulation Initiative and a researcher at Stanford University’s Center for International Security and Cooperation, noted that actors tend to accumulate errors by placing too much trust in AI conclusions, a tool often designed to minimize uncertainty while showing a tendency toward escalation and more “hallucinations” than humans. The researcher added that, in most simulations, actors who overestimate the military potential of opposing AI systems showed a greater proclivity to strike first, targeting especially enemy nuclear forces .
This shift toward human factors also applies to the information front. If the so-called slopaganda worries, it is not because it would present a coherent alternative narrative, but because it erodes reality precisely by its raison d’être: dissolving it. Whereas Soviet propaganda aimed to create a shared confidence in a radiant future, its digital heir thrives in the blank of a news feed, where each assertion is disconnected from its source and its consequences. The weapon is no longer the lie but the generalized doubt: a terrain on which Russia, unable to compete technologically, retains a clear comparative advantage.
Ultimately, the true “revolution in the art of war” sparked by artificial intelligence may hinge less on computational power and battlefield data assessment than on entirely human factors: the sense of vulnerability and the trust in the quasi-divine instruments we have crafted with our own hands.