Purportedly AI-Generated Airport Videos Reportedly Showed Diaspora Voters Supporting Albin Kurti

Facebook pages are reported to have posted purportedly AI-generated airport-style interview videos falsely showing expatriates saying they would return on December 28 to vote for Albin Kurti. The videos reportedly used Pristina Airport settings and were assessed by fact-checkers' verification tools as AI-generated or deepfake content.

Risk classification

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

The incident involved the coordinated creation and dissemination of AI-generated deepfake videos on social media to manipulate public opinion and influence the Kosovo parliamentary elections.

Additional risk subdomains

• 3.1 False or misleading information: The AI-generated deepfakes fabricated statements and interviews to spread false information about voter intentions.

Causal factors

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

The incident was caused by human actors intentionally using deployed AI generation tools to create and distribute deepfake videos to manipulate public opinion ahead of the Kosovo elections.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The incident involves 'AI-generated content such as deepfakes' which, under the EU AI Act, require specific transparency obligations to ensure users are informed.

AI system and alleged parties

• AI system: unspecified
• AI purpose: Deepfake Video Generation; Voice Generation
• Behaviour type: Tool
• Alleged developer: Voice synthesis technology developers, Deepfake technology developers
• Alleged deployer: Information manipulation actors, Deepfake creators
• Alleged harmed parties: National security and intelligence stakeholders, General public of Kosovo, General public, Epistemic integrity, Democratic integrity, Albin Kurti

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. 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 Minor, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Minor, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: The report explicitly describes the creation and spread of malicious deepfake videos designed to deceive the electorate.

Directly caused: AI-generated videos with fabricated audio and visuals were spread on Facebook and TikTok, accumulating over 10,600 views to manipulate public opinion.

Inferred additional harm: It is likely that additional AI-generated videos and audio clips were spread across other social media channels, reaching tens of thousands of voters.

Democracy

Reported: The report explicitly describes attempts to manipulate public opinion and influence voter behavior ahead of the December 28 parliamentary elections.

Directly caused: Deepfake videos fabricated diaspora support for Albin Kurti, directly targeting the integrity of the democratic election process in Kosovo.

Inferred additional harm: The widespread use of such deepfakes could erode public trust in the election results, democratic institutions, and the authenticity of political discourse.

Epistemic

Reported: Yes, the report details the spread of AI-driven misinformation and fabricated video/audio content.

Directly caused: AI-generated videos fabricated interviews with non-existent citizens expressing political opinions, creating confusion and spreading false narratives.

Inferred additional harm: The proliferation of highly realistic deepfakes makes it harder for citizens to distinguish between real and fabricated news, degrading the overall information ecosystem.

People affected

• Occurrences reported: 1
• People reportedly exposed: 10600

Potential causes

Management

• Weak Platform Governance: Social media platforms did not implement safeguards against election deepfakes.

Technology

• AI Deepfake Generation Tools: Tools allowed creation of realistic fake videos and audio of citizens.
• Visual and Audio Distortions: AI generation left visible distortions in text and synthetic audio patterns.

Data Inputs

• Synthetic Audio Training Data: AI models used voice data to synthesize false statements by US analysts.
• Disconnected Media Contexts: Existing images and documents were combined with AI audio to deceive.

Human Factors

• Vulnerability to Deception: Audience easily deceived by fake visual and audio content during elections.
• Intentional Political Manipulation: Actors intentionally used AI to manipulate public opinion and support Kurti.

Process and Methods

• Inadequate Platform Moderation: Facebook and TikTok failed to proactively detect and block deepfakes.
• Post-Facto Verification Reliance: Detection relied on external fact-checkers like hibrid.info after spread.

Regulatory Environment

• Lack of AI Content Regulations: Absence of strict rules governing AI-generated political content in Kosovo.

Information quality

• Classification confidence: High
• Reason for confidence: The reports from hibrid.info and kallxo.com provide clear, verified details about the specific Facebook pages, video content, view counts, and the fact-checking tools (Hive Moderation, deepware.ai, WasitAi) used to confirm the AI generation. There is little ambiguity regarding the nature of the deepfakes or their intended political impact.
• Ambiguities identified: The specific AI generation tools used to create the deepfakes are not identified, only the detection tools are named.
• Alternative interpretations: None. The evidence clearly points to a coordinated political disinformation campaign using synthetic media.

Ahead of the December 2025 Kosovo elections, malicious actors deployed highly realistic AI-generated deepfakes depicting diaspora support for Albin Kurti. While the immediate scale was limited and the content was debunked, the incident represents a direct attempt to manipulate democratic processes and sovereignty through synthetic media.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: Kosovo
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Near-term. The incident occurred in the immediate run-up to the December elections, representing an active electoral threat requiring monitoring.
• Autonomy: Human-controlled. AI tools were used strictly as instruments by human actors to generate specific content, with no autonomous decision-making.
• Novelty: Evolved capability. While political deepfakes are an established concept, the high level of realism represents an evolved capability.

Impact by dimension

• Physical security: Negligible. No physical attacks, threats to critical infrastructure, or kinetic safety risks were reported in connection with this incident.
• Information security: Minor. Coordinated disinformation campaign utilized highly realistic AI-generated video and audio to target political leadership, though the reach was limited and quickly debunked by fact-checkers.
• Sovereignty: Substantial. The incident represents a direct attempt to manipulate public opinion and influence the outcome of Kosovo's parliamentary elections, threatening democratic processes.
• Economic security: Negligible. No economic damage, financial loss, or strategic technology theft was reported.
• Societal stability: Minor. The proliferation of political deepfakes risks eroding public trust in democratic institutions and degrading the overall information ecosystem, though no civil unrest resulted.