Russia Reportedly Using Artificial Intelligence in Disinformation Campaigns to Erode Western Support for Ukraine

The Russian government has been stepping up its foreign influence campaigns by using artificial intelligence and emerging technologies to spread disinformation and sow distrust in policies supportive of Ukraine. Part of the strategy includes carrying out influence laundering operations by disseminating their messages to the American public via allies inside nominally independent organizations, according to a recent declassified analysis.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.1 Disinformation, surveillance, and influence at scale

The incident involves coordinated, state-sponsored campaigns using AI-driven bot networks and deepfakes to conduct large-scale disinformation and manipulate public opinion regarding the war in Ukraine.

Additional risk subdomains

• 3.2 Pollution of information ecosystem and loss of consensus reality: The massive influx of automated bot activity and synthetic media pollutes social media platforms, degrading the shared information ecosystem.
• 1.2 Exposure to toxic content: Coordinated networks of fake accounts directed online harassment and intimidation at citizens expressing support for Ukraine.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The disinformation campaign and deployment of bot networks were intentionally directed by human actors (the Russian government and intelligence services) to influence public opinion.

EU AI Act risk tier

• Risk tier: 1 Unacceptable

Unacceptable Risk: The reports describe the use of AI-driven bot networks and deepfakes for large-scale behavioral influence and cognitive manipulation of public opinion, which poses significant dangers to society and democratic processes.

AI system and alleged parties

• AI system: Botometer, unspecified
• AI purpose: Social Media Content Generation; Content Recommendation
• Behaviour type: Multi-agent
• Alleged developer: Russian government
• Alleged deployer: Russian government, Misinformation spreaders, FSB, Federal Security Service, Disinformation spreaders
• Alleged harmed parties: Ukraine, Truth, National security and intelligence stakeholders, Media integrity, Journalistic integrity, General public of the European Union, General public, Epistemic integrity, Electoral integrity, Democracy, American public

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Substantial. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Substantial, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Minor, indirect Negligible
• Democracy: direct Minor, indirect Negligible
• Privacy: direct Minor, indirect Negligible
• Psychological: direct Minor, indirect Negligible
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: Yes, the reports describe the creation and spread of toxic, malicious, and harassing content.

Directly caused: Coordinated networks of bots and fake accounts spread pro-Russian propaganda, conspiracy theories, and directed online harassment at Bulgarian citizens supporting Ukraine.

Inferred additional harm: Millions of social media users were exposed to toxic and polarizing political content designed to sow discord and fear.

Civil rights

Reported: Yes, the reports describe violations of the right to free expression and freedom from intimidation.

Directly caused: Coordinated harassment and intimidation campaigns targeted Bulgarian citizens to suppress their public support for Ukraine.

Inferred additional harm: Widespread automated harassment likely chilled public discourse and violated the civil rights of thousands of citizens in Eastern Europe by intimidating them into silence.

Democracy

Reported: Yes, the reports explicitly describe efforts to influence democratic elections and political processes.

Directly caused: Russian intelligence agencies used propaganda and conspiracy theories to end U.S. and European support for Ukraine, aiming to influence political debates, congressional spending, and European parliamentary elections.

Inferred additional harm: The systematic manipulation of public discourse through automated bot networks and AI-generated content poses a long-term threat to the integrity of democratic elections and public trust in democratic institutions across Western nations.

Privacy

Reported: Yes, the reports mention hacking of social media accounts.

Directly caused: The 'Ghostwriter' campaign hacked the social media accounts of European political figures and news outlets to spread fabricated content.

Inferred additional harm: It is likely that further unauthorized access and credential theft occurred to compromise accounts for disinformation dissemination, though specific numbers are not detailed.

Psychological

Reported: Yes, the reports describe online harassment and intimidation directed at individuals.

Directly caused: Report 1 explicitly notes that anonymous accounts directed online harassment and coordinated intimidation tactics at Bulgarians who expressed support for Ukraine.

Inferred additional harm: It is likely that thousands of social media users supporting Ukraine experienced distress, anxiety, or fear due to coordinated bot-driven harassment campaigns, though specific numbers are not quantified.

Epistemic

Reported: Yes, the reports describe extensive AI-driven misinformation, deepfakes, and conspiracy theories.

Directly caused: Deepfakes of Zelenskyy and Putin were circulated to spread fabricated peace declarations, and fake accounts spread debunked conspiracy theories claiming the U.S. runs secret biolabs in Ukraine.

Inferred additional harm: The massive influx of automated disinformation has eroded shared consensus reality for millions of social media users, making it increasingly difficult to distinguish real news from synthetic propaganda.

People affected

• Occurrences reported: 1
• People reportedly exposed: 100000000

Potential causes

Management

• Dismantling Safety Teams: Platforms like X dismantled teams hunting for election interference.
• Prioritizing Engagement over Safety: Platform algorithms prioritize viral content over factual accuracy.
• Inadequate Risk Assessment: Management failed to anticipate the scale of bot networks using open APIs.

Technology

• AI-Enabled Bot Networks: Automated bot accounts rapidly amplify Russian disinformation and hashtags.
• Generative AI and Deepfakes: Generative tools create realistic fake videos and texts at near-zero cost.
• Recommendation Algorithms: Platform algorithms prioritize viral content over factual accuracy.

Data Inputs

• Open Platform APIs: Flexible APIs allow bad actors to automate accounts and scrape data easily.
• Biased Training Data: LLMs trained on Western data contradict Russian state-approved narratives.
• Lack of Proactive Detection Data: AI detection models struggle with novel disinformation without human labeling.

Human Factors

• High Susceptibility to Deepfakes: Users trust manipulated media like fabricated videos of political leaders.
• User Polarization: Polarized audiences readily accept and amplify unverified conspiracy theories.
• Mass Flight of IT Talent: Brain drain of Russian tech experts limits domestic AI safety development.

Process and Methods

• Inadequate Automated Moderation: Platforms fail to proactively identify and remove automated bot networks.
• Government Account Loophole: Rules restricting state media do not apply to official embassy accounts.
• Fragmented Platform Policies: Inconsistent moderation across platforms creates safe havens for propaganda.

Regulatory Environment

• Fragmented Global Regulations: Divergent AI and data laws make coordinated global moderation difficult.
• Punitive Disinformation Laws: Strict domestic laws force Russian AI models to self-censor on key events.
• Western Tech Sanctions: Restrictions on hardware limit the computing power needed for safe AI models.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide extensive, multi-perspective coverage of the disinformation campaigns, including academic research using Botometer, official statements from tech platforms like Twitter, and intelligence assessments from U.S. officials. The role of AI in generating deepfakes and coordinating bot networks is clearly documented with specific metrics (e.g., 75,000 accounts removed, Botometer scores).
• Ambiguities identified: The exact proportion of disinformation generated by AI versus human trolls is not fully quantified, and the precise algorithms used by platforms to recommend content are proprietary.
• Alternative interpretations: Some platforms, like Facebook in Report 1, argued that certain flagged accounts were operated by single human users rather than coordinated bot networks, suggesting some platform manipulation may be organic rather than automated.

Russia conducted large-scale, state-sponsored information operations using AI-driven bot networks, deepfakes, and automated amplification to manipulate public opinion and erode Western support for Ukraine. The incident represents a severe information warfare threat and a substantial challenge to democratic sovereignty and societal stability.

• Overall national security impact: Severe
• Scope: Multiple nations
• Primary target: Ukraine and Western allies
• Other affected: USA, Eastern Europe, Global South
• Alleged perpetrator: Russia

Threat characteristics

• Imminence: Long-term. Represents an ongoing, persistent strategic threat and capability rather than a sudden, localized 72-hour crisis.
• Autonomy: Human-supervised. AI tools, bot networks, and generative models are deployed and directed by human operators to achieve specific strategic objectives.
• Novelty: Evolved capability. Represents a significant advancement of existing information warfare tactics through the integration of generative AI and automated bot networks.

Impact by dimension

• Physical security: Negligible. No physical attacks, critical infrastructure manipulation, or kinetic targeting occurred during these information operations.
• Information security: Severe. State-sponsored information warfare used AI-generated deepfakes of heads of state and massive automated bot networks to manipulate public opinion and influence defense policies.
• Sovereignty: Substantial. Coordinated campaigns targeted democratic decision-making, congressional spending debates, and European elections to undermine state policies and support for Ukraine.
• Economic security: Minor. Sanctions impacted Russian tech firms like Yandex and Sberbank, but the incident itself did not involve direct theft of strategic Western AI models or financial system attacks.
• Societal stability: Substantial. Automated harassment targeted citizens expressing support for Ukraine, while widespread synthetic propaganda aimed to polarize public discourse and erode trust in democratic institutions.