On June 6, 1972, a week after President Richard Nixon concluded his historic visit to Moscow, the Soviet Union premiered the film Residence Permit. Ostensibly a simple drama, the movie was in fact a meticulously crafted piece of state propaganda. It told the story of a Leningrad doctor who, in pursuit of perceived Western freedoms and professional advancement, chose to remain in Western Europe. His journey, however, led him to a harsh realization: the narratives of Western prosperity were a fallacy, and his decision to leave the Soviet Union, the film concluded, was a grave error. To disseminate this message, the Soviet Union marshalled an extensive apparatus: film studios, a pervasive censorship board, vast distribution networks, and influential cultural institutions. The infrastructure required for such propaganda was colossal, immensely costly, and ponderously slow. Over five decades later, Russia’s fundamental objectives remain strikingly similar. It continues to aim at projecting specific narratives, both internally and on the international stage. What has dramatically transformed, however, is the cost, speed, and sheer scale with which these ambitions can now be pursued.
Today, the Kremlin can achieve in mere minutes what once necessitated months of coordinated effort. Furthermore, these efforts can be executed across dozens of languages, disseminated through hundreds of digital platforms, and deployed at a scale that no Soviet-era propagandist could have possibly envisioned. The catalyst for this transformation is artificial intelligence. However, this is not the AI typically discussed in the context of a global technological arms race, which usually refers to the competition to develop the most powerful AI models, the fastest microprocessors, and the most sophisticated autonomous systems. By that metric, Russia occupies a complex and somewhat precarious position, primarily due to a significant structural constraint: hardware.
Samuel Bendett, a seasoned advisor to the Russia Studies Programme at CNA, a prominent think tank based in the Washington, D.C. metropolitan area, observes that Russia possesses a robust reservoir of talent, comprising STEM-educated specialists and mathematicians who are demonstrably capable of developing advanced software. "However," Bendett notes, "hardware has consistently been the Achilles’ heel, a weakness that can be traced back to the early days of the Cold War." This hardware deficit carries substantial weight in the contemporary AI landscape. State-of-the-art machine learning systems are critically dependent on specialized chips, such as graphics processing units (GPUs) and AI accelerators, which are engineered to perform an immense volume of mathematical computations concurrently.
Currently, Russia faces an insurmountable challenge in domestically producing the advanced semiconductors essential for frontier AI development. The imposition of Western sanctions following its invasion of Ukraine has exacerbated these difficulties, creating a cascade of new problems. Consequently, Moscow is compelled to rely on smuggled components or those sourced from China for its more sophisticated technological systems. "Russia has a strong reliance on NVIDIA microchips, particularly for military-related AI applications. The same dependency extends to hardware such as Raspberry Pi and Orange Pi," Bendett elaborates. "This critical hardware is not manufactured within Russia, or if comparable domestic equivalents exist, they are already considerably outdated when measured against global standards."
Despite these significant hardware limitations, they have had a negligible impact on curbing the Kremlin’s broader strategic objectives. Instead, Russia has strategically pivoted its focus to a different kind of operational theatre, one where semiconductor shortages are far less consequential. In this emerging domain, the paramount objective is not the development of the most advanced AI systems, but rather the meticulous shaping of the very environment in which these systems operate. This encompasses influencing the information retrieved by Western AI models, maintaining stringent control over the information accessible to its own populace, and dictating the perceptions of audiences beyond its borders.
Projecting Influence Beyond Borders
Sopo Gelava, who has dedicated over a decade to researching disinformation and has been associated with the Atlantic Council’s Digital Forensic Research Lab since 2020, observes a significant intensification in the use of AI within Russian disinformation campaigns in recent years. "Actors who previously engaged in manual content creation now exhibit far less direct human involvement," she states. Gelava explains that even a single operation, originating from a Russian website and subsequently propagating across multiple platforms in various languages, can exhibit clear indicators of AI utilization throughout its entire lifecycle.
"Either automation is being employed, or AI is actively involved in generating the content," Gelava elaborates. "While this hasn’t necessarily instigated a revolutionary paradigm shift in the realm of disinformation, it has undeniably rendered it far more scalable. It bestows upon creators significantly greater capacity to disseminate content at unprecedented speeds, enabling them to reach exceedingly large audiences. Overall, AI empowers them to achieve a substantially greater impact." These influence operations are typically most active in nations where Moscow has discernible political interests. They tend to escalate in intensity in the periods preceding elections but do not cease once voting concludes; the underlying narratives persist, adeptly adapting to new events and evolving audience demographics.
In a notable recent instance, the Digital Forensic Research Lab identified a coordinated network of TikTok accounts. These accounts appeared to be orchestrating the dissemination of AI-generated content specifically targeting Moldova’s ruling Party of Action and Solidarity and its President, Maia Sandu, while simultaneously encouraging individuals to participate in protests. Gelava further details that AI is deployed in a multifaceted manner within these operations. It can automate the synchronized propagation of narratives across diverse platforms or enhance visual content to amplify emotional resonance and thereby increase user engagement. The ultimate objective, however, remains remarkably consistent: to reach as many individuals as possible, with the utmost efficiency.
In the specific case of Moldova, at the time of this analysis, the aggregated following across the scrutinized TikTok accounts surpassed 158,556 users, with total engagement exceeding 26.3 million interactions across all types. AI is not only facilitating the scaling of disinformation but is also intensifying cyber warfare capabilities. In April, Dutch military intelligence issued a warning that Russia is actively employing AI to accelerate cyberattacks, a threat that is projected to escalate further. The evolving landscape is characterized not merely by the increased speed of these operations but also by their structural metamorphosis. AI is transitioning cyberattacks from labor-intensive endeavors to highly automated processes, thereby enabling the simultaneous identification and targeting of multiple entities. Tasks that once demanded sustained human effort can now be executed in mere seconds, dramatically expanding both the scope and the reach of these sophisticated operations.
Domestic Information Control
The underlying principles driving Russia’s application of AI abroad—namely, automation, scalability, and efficiency—are increasingly being mirrored and implemented within its domestic borders. "Internal security has always been a paramount concern and a primary driver of Russia’s high-tech development in general," states Bendett. "A key objective has been to insulate the nation from external influence and to mitigate its impact on the domestic population." AI is now rendering this strategic approach exponentially more potent. Where state propaganda once relied on an extensive physical infrastructure comprising studios, printing presses, and distribution networks, it can now be managed entirely through digital means. This empowers the Kremlin to monitor, filter, and meticulously shape its information environment with significantly fewer human resources and at a vastly expanded scale.
In January, Forbes reported that Roskomnadzor, Russia’s federal agency responsible for regulating and censoring telecommunications, intends to deploy a machine learning-based system for filtering internet traffic within the next year. According to the agency’s digitalization plan submitted to the government, a substantial allocation of 2.27 billion rubles (approximately $30 million) has been earmarked for this initiative. Media reports indicate that the system aims to identify and block prohibited content with enhanced efficiency and to restrict access to Virtual Private Network (VPN) services, which Russian citizens utilize to circumvent existing censorship measures.
Bendett posits that the current trajectory of restrictions on platforms like Telegram, broader internet blocking, and limitations on VPN usage runs counter to long-term strategic logic. He argues that severing the access of a significant portion of the Russian populace to international IT applications and global messaging platforms could hinder future development, particularly given the relatively nascent state of Russia’s IT and high-tech sector. "Russia’s government policies, which are currently focused on curtailing the population’s access to some of these international components, are likely a self-inflicted wound," Bendett suggests. "This approach is delaying numerous projects and advancements that would otherwise have unfolded had Russian developers and users maintained access to Western applications, databases, and algorithms."
The sophisticated surveillance architecture is also extending its reach into the physical realm. Across Russian cities, street cameras equipped with AI-powered recognition systems are being deployed to monitor public spaces and identify individuals in real-time. In Yekaterinburg alone, an estimated 1,000 additional cameras are slated for installation by the end of June, comprehensively covering streets and public areas. These systems continuously analyze video feeds, thereby significantly augmenting the state’s capacity to monitor its population without necessitating a proportional increase in human personnel.
Subtly Steering AI’s Learning Process
Russia’s operational endeavors, both on the international stage and within its domestic sphere, are largely discernible, even if challenging to counteract. However, there exists another dimension to its AI strategy that is considerably more elusive and difficult to detect. Recent academic studies suggest that one of Russia’s most impactful AI strategies is not directly targeting populations but rather the AI models that an increasing number of people rely upon to interpret and comprehend the world around them. This strategic approach aims to subtly influence Western AI models by shaping the very data they are trained on and the sources from which they retrieve information.
A network of pro-Kremlin websites has reportedly been instrumental in flooding the internet with millions of pieces of Russian propaganda, utilizing AI tools to amplify their reach. A significant portion of this content is meticulously designed to be readily ingested by search engines and subsequently scraped into the large datasets that are fundamental for training AI systems. Researchers have characterized this tactic as a form of "data poisoning by scale," wherein the objective is not to disseminate a single piece of misinformation but rather to achieve a sustained saturation of the digital information ecosystem. The overarching concern is that, over time, this sustained saturation could subtly influence how AI systems interpret, prioritize, and ultimately reproduce information, thereby embedding specific biases within their core functionalities.
Sopo Gelava highlights that Russia’s AI-driven tactics are progressively becoming more sophisticated. AI-generated content that was once prevalent in disinformation campaigns was relatively easy to identify. However, this situation is rapidly evolving. "There were frequent grammatical errors, and in the past, we could often discern from these errors that the operation had been created by AI. Today, however, it offers more opportunities to the creators of disinformation because the translation is much more refined and significantly better adapted to the local context," Gelava observes. Researchers who are closely studying Russia’s utilization of AI believe that its parallel efforts in the global AI race are becoming increasingly difficult to detect, more scalable, and more precisely targeted. These sophisticated operations not only reach broader audiences but also pose a significant risk of fundamentally shaping how information is interpreted and reproduced across a wide array of digital systems, potentially influencing global understanding and decision-making.
