The Double Edge of Russian Sovereign AI: Isolation and Narrative Consistency

The rapid development of commercial-level artificial intelligence (AI) has lowered the costs of sophisticated, large-scale information operations, including foreign information manipulation and interference (FIMI) campaigns. At its core, generative AI—used to create multimedia content from user prompts, called ‘synthetic’ content—is being deployed ‘off the shelf’ by threat actors worldwide to deceive and persuade targets at scale.

Due to their scale and ambition, Russian FIMI campaigns have become a feature of international politics. Russian attempts to sway elections and polarise society in NATO Allies, including France and the US, as well as countries worldwide, such as Côte d’Ivoire and Japan, have become routine. According to the 2026 4^th European External Action Service (EEAS) report on FIMI, 29% of all FIMI campaigns documented by the service in 2025 can be attributed to Russia. The report also found that 27% of all FIMI incidents documented involved AI-enhanced content (e.g., synthetic content).^[1]

Implementation

Guiding the growing use of generative AI by Russian actors is the notion of ‘information confrontation’ (информационное противоборство), a narrative that frames international politics as a realm of conflict, playing out in the information domain.^[2] Russian officials and thinkers are reflecting on how to marshal this tool to the country’s advantage in this confrontation.

The Russian Intelligence Studies journal “Невидимое измерение” (An Invisible Dimension) has widely covered the use of AI across the defence and security sectors. In a 2024 article, frequent contributor Sergey Denisentsev presents several generative AI functions of interest to intelligence services, namely: automated data gathering, data mining, analysis of biometric data, support for hacking, production of deepfakes, social engineering through disinformation and blackmail, spam, and bot network creation.^[3]

Since the 2022 launch of ChatGPT’s first commercial model, AI-enhanced FIMI campaigns launched by Russian state-backed and state-aligned actors have been recurrent.

Since the 2022 launch of ChatGPT’s first commercial model, AI-enhanced FIMI campaigns launched by Russian state-backed and state-aligned actors have been recurrent. According to the Microsoft Threat Analysis Center (MTAC), the US 2024 presidential election was targeted by Russian state-linked groups using generative AI.^[4] The MTAC team drew several conclusions about how Russian threat actors deployed AI: AI-generated audio was more impactful than video; AI-enhanced—rather than AI-generated—video was more commonly used; dissemination around pivotal events and in closed groups had the most impact.

Despite the relevance of these Russian efforts, MTAC concluded that the hypothesis that AI-generated deepfakes could create “mass deception” ultimately “did not bear out.” The EEAS 2026 report found that synthetic content was being deployed on a large scale but with low impact, due to attackers’ high-volume, low-quality approach to FIMI campaigns.

The other side appears to concur with this nuanced assessment. A 2025 Royal United Services Institute (RUSI) report found Russian hacker groups boasting of their proficiency in using generative AI for FIMI production. In turn, operational and technical competence, as well as strategic direction, are seen by some Russian state-aligned hacker groups as essential for AI-enhanced FIMI campaigns not to “do more harm than good.”^[5]

From Global Leader to Isolation

The frustration expressed in the RUSI report reflects the entrenched anxiety in Russia’s ruling circles about the asymmetry in the AI sector, namely, between domestic and western companies. Owing to lacklustre results and the fallout from Russia’s full-scale war of aggression against Ukraine, the evolution of state AI policy has gone from seeking leadership to falling into isolation. This development has consequences for AI-enhanced Russian FIMI.

The evolution of state AI policy has gone from seeking leadership to falling into isolation.

In 2019, Russian ruler Vladimir Putin delivered a speech on the future of AI that Russian sources consider a pivotal moment in state policy in the sector.^[6] In that speech, Putin referred to AI as a “resource of colossal power” and a national cause: “We must, and I am confident that we can become one of the global leaders in AI. This is a matter of our future, of Russia’s place in the world.”^[7] That same year, the Russian national AI strategy was published, with a time horizon to 2030.^[8]

As of 2026, the results are mixed. International sanctions, especially those targeting electronics, and the broader isolation of Russia’s technology sector have hindered the Kremlin’s AI goals.^[9] Another restraining factor is the structure of the technology sector in Russia, dominated by large state-owned or state-connected companies, with relatively little scope for the development of a start-up ecosystem driving innovation. The post-2022 militarisation and economic stagnation compound these weaknesses. In turn, state resources and the relative competitiveness of Russian universities have kept ambitions afloat.^[10] According to the aggregator DisserCat, 279 doctoral theses have been registered at Russian universities on AI since 2023.^[11]

The Kremlin has increasingly framed the AI sector in terms of ‘sovereignty’ and national security.

Today, rather than developing AI solutions that can compete with products such as ChatGPT, Mistral AI, or Anthropic’s Claude, the Kremlin has increasingly framed the country’s AI sector in terms of ‘sovereignty’ and national security. Namely, the state goal has shifted towards developing local generative AI solutions and isolating these models from ‘western’ information. Russian analysts have been pondering this topic for years. For example, one 2021 article compares Chinese and North Korean models of AI ‘data filtering’ to draw conclusions about how Russia could proceed with its own model.^[12] Reflecting a paranoid view of information from abroad, as of April 2026, the Kremlin has been drafting a law proposal requiring that Russian AI models be trained only on data generated in Russia.^[13]

The Kremlin is not the only one concerned with the ideological rigour of Russian AI models. The aforementioned RUSI study found hacker groups express their frustration with Russian generative AI solutions, such as Yandex Alisa and Sber’s GigaChat. The complaint highlighted by the report was the lack of ideological consistency in these models, such as sometimes not referring to the temporarily occupied territories of Ukraine as “Russian.” As one Ukrainian journalist reported, Russian AI “knows the truth. It is programmed to hide it.”^[14]

Outlook

In the past, Russia state and state-linked actors have exploited major opportunities to leverage technological breakthroughs for political gain. Reportedly, Russian state-linked actors knew about the Cambridge Analytica efforts as early as 2015, well before these were detected by US officials or the wider public after the FIMI campaign targeting the 2016 US presidential election.^[15] Like the most impactful AI FIMI campaigns, Cambridge Analytica also enabled micro-targeting.

Russia can gain a strategic advantage from cutting-edge technologies, even when they are not domestically produced.

As the Cambridge Analytica case suggests, Russia can gain a strategic advantage from cutting-edge technologies, even when they are not domestically produced. The implementation of foreign generative AI tools might eventually lead to indigenisation of the technology and the development of more sophisticated attack methodologies using international AI models. While the development of ‘sovereign AI’ appears contradictory, it might have advantages for Russian FIMI campaigns. Namely, a more ideologically consistent Russian AI tools and services could also facilitate the creation of synthetic content that, while being adaptive to requests, stays in line with Kremlin propaganda.

Though facilitated by low costs, the results are mitigated by a focus on quantity over quality. Still, Russian state policy on AI is driven by both a sense of threat and opportunity, with isolation reflecting a state policy increasingly focused on retrenchment, regime survival, and its war of aggression against Ukraine.^[16] Despite these and other limitations, persistence and resource allocation, as well as high volume, can create laboratory conditions where experimentation results in new impactful attack methodologies.

^[1] “4th EEAS Annual Report on Foreign Information Manipulation and Interference Threats,” European External Action Service, March 2026, 10.

^[2] Karen-Anna Eggen, “A Strategy for the Weak: The Role of Information Confrontation in Russia’s Grand Strategy,” Defence Studies, 2025, 1–25.

^[3] Sergey Denisentsev, “Перспективы применения искусственного интеллекта в разведывательной работе [Prospects for the application of artificial intelligence in intelligence work],” Nevidmoye Izmerenie 1, no. 2, 2023.

^[4] “Nation-states engage in US-focused influence operations ahead of US presidential election,” MTAC Report, 17 April 2024.

^[5] Antonio Giustozzi, “Can AI help Russia Decisively Improve its Information War Against the West?,” Royal United Services Institute, 27 June 2025.

^[6] Lapshin, V. F., “Artificial intelligence technology as a potential threat to public security protected by criminal law,” Russian Journal of Deviant Behavior 2(4) (2022): 374–385.

^[7] President of Russia, “Conference on artificial intelligence,” Kremlin.ru, 9 November 2019. 

^[8] President of Russia, “О развитии искусственной интеллекта,” Kremlin.ru, 10 October 2019.

^[9] “В «Роснефти» заявили о невозможности создать по требованию Путина «суверенный» ИИ [Rosneft says it is impossible to create a ‘sovereign’ AI as Putin demanded],” The Moscow Times, 17 April 2026. 

^[10] Heidi Hanhijärvi, “Artificial Intelligence and Foreign Information Manipulation: Chinese and Russian approaches,” The European Centre of Excellence for Countering Hybrid Threats, Hybrid CoE Paper 29, March 2026.

^[11] Figure as of April 2026. Dissercat website.

^[12] I. Dzyaloshinsky, “Communication Services in the Age of Digital Civilization: Convergence with Artificial Intelligence,” Communication. Media. Design, vol 6 no. 3 (2021).

^[13] Dariia Mykhailenko, “Inside Russia’s “Sovereign AI” Plan—And Why It May Not Work,” United 24 Media, 17 April 2026.

^[14] Ihor Samokhodskyi, “I tested Russia’s AI. It knows the truth, but it’s been trained to lie,” The Kyiv Independent, 5 February 2026.

^[15] Carole Cadwalladr and Emma Graham-Harrison, “Cambridge Analytica: links to Moscow oil firm and St Petersburg university,” The Guardian, 17 March 2018.

^[16] Ivan U. Klyszcz, “Prepare for Russia’s Coming Retrenchment,” PONARS Eurasia Policy Memo, 20 November 2022.