Deepfake Video of Ukrainian President Yielding to Russia Posted on Ukrainian Websites and Social Media

A quickly-debunked deepfaked video of the Ukrainian President Volodymyr Zelenskyy was posted on various Ukrainian websites and social media platforms encouraging Ukrainians to surrender to Russian forces during the Russia-Ukraine war.

Risk classification

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

The deepfake was deployed as part of a coordinated disinformation and propaganda campaign aimed at manipulating public opinion and population-level behavior during a war.

Additional risk subdomains

• 3.2 Pollution of information ecosystem and loss of consensus reality: Experts noted that the incident pollutes the information ecosystem and contributes to the 'liar's dividend,' making it harder for the public to trust authentic media.

Causal factors

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

The incident was caused by human actors (hackers and propagandists) who intentionally created and distributed the deepfake video to deceive the public during an armed conflict.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The report describes the creation and dissemination of a deepfake video, which is explicitly classified under the EU AI Act as a limited risk system requiring transparency obligations so users know they are interacting with AI-generated content.

AI system and alleged parties

• AI system: unspecified
• AI purpose: Deepfake Video Generation; Voice Generation
• Behaviour type: Tool
• Alleged developer: unknown
• Alleged deployer: hackers
• Alleged harmed parties: Volodymyr Zelenskyy, Ukrainian social media users, Ukrainian public

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

Infrastructure

Reported: Yes

Indirectly caused: The live broadcast news ticker of Ukraine 24 and its website were compromised by hackers to display the deepfake message.

Malicious content

Reported: Yes

Directly caused: The AI system was used to generate a deepfake video of President Zelenskyy falsely calling for Ukrainian forces to surrender.

Indirectly caused: The deepfake video was uploaded to hacked websites, broadcast on TV, and shared widely across social media platforms including Facebook, YouTube, Twitter, and VKontakte.

Democracy

Reported: Yes

Indirectly caused: The deepfake was used as a weapon of information warfare during an active armed conflict, attempting to undermine the authority of a democratically elected leader and manipulate public opinion.

Epistemic

Reported: Yes

Directly caused: The AI system generated a highly deceptive (though poorly executed) video of President Zelenskyy making a false surrender statement.

Indirectly caused: The dissemination of the deepfake on social media and hacked news broadcasts created temporary confusion and threatened to erode public trust in authentic video communications from political leaders.

Inferred additional harm: The incident contributes to the 'liar's dividend,' making it easier for public figures to deny authentic footage and increasing general public skepticism toward real news.

People affected

• Occurrences reported: 1
• People reportedly exposed: 100000

Potential causes

Management

• State-Sponsored Disinformation: Adversaries integrated AI deepfakes into their broader military strategy.

Technology

• Accessible Deepfake Software: Faceswap and AI voice synthesis tools are readily available to adversaries.
• Low-Resolution Masking: Low quality was used intentionally to hide rendering glitches in the video.
• Stilted AI Voice Synthesis: AI voice cloning generated a robotic but passable audio track of the leader.

Data Inputs

• Abundant Source Material: High volume of public video of the target was used to train the AI model.
• Hacked Broadcast Feed: Hackers injected fake text into the news ticker to support the deepfake.

Human Factors

• Cognitive Fog of War: High-stress conflict makes the public more susceptible to panic and lies.
• The Liar's Dividend: The existence of fakes makes citizens doubt the authenticity of real media.

Process and Methods

• Vulnerable News Infrastructure: Lack of robust security allowed hackers to breach the TV station's systems.
• Reactive Content Moderation: Social media platforms removed the video only after it started spreading.

Regulatory Environment

• Inadequate Platform Mandates: No regulatory standards exist to force proactive deepfake identification.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide consistent, detailed, and well-sourced accounts of the incident from multiple reputable news organizations. The facts regarding the creation, distribution, hacking of Ukraine 24, and subsequent debunking of the deepfake are clear and uncontradicted.
• Ambiguities identified: The exact AI software or model used to generate the deepfake is not identified, and the specific actors behind the creation of the video remain unconfirmed, though widely suspected to be Russian state-sponsored or aligned hackers.
• Alternative interpretations: None. The event is universally recognized as an attempted political disinformation campaign using deepfake technology.

A deepfake video of Ukrainian President Zelenskyy falsely calling for surrender was broadcast via a hacked news channel and social media. While the low quality and rapid debunking limited its immediate impact, it represents a first-of-its-kind use of AI-generated media as an information warfare weapon during active armed conflict.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: Ukraine
• Alleged perpetrator: Russian-aligned hackers

Threat characteristics

• Imminence: Immediate. The incident unfolded in real-time during an active military invasion, requiring immediate, coordinated counter-disinformation responses within hours.
• Autonomy: Human-controlled. The AI system functioned strictly as a human-controlled tool to generate the synthetic media, with humans managing the creation, hacking, and distribution.
• Novelty: First-of-its-kind. This was the first major documented instance of a deepfake of a state leader being deployed as an information weapon during an active armed conflict.

Impact by dimension

• Physical security: Minor. Hackers compromised Ukraine 24's live broadcast ticker to distribute the deepfake, representing a minor breach of media infrastructure but causing no physical damage.
• Information security: Substantial. The incident represents a notable information warfare operation during active conflict, using a deepfake of a head of state to attempt to induce military surrender, though its low quality limited its impact.
• Sovereignty: Substantial. The deepfake directly targeted the authority of Ukraine's democratically elected president to undermine state resistance, though the swift official debunking prevented disruption to government functions.
• Economic security: Negligible. No significant economic manipulation, technological theft, or threat to strategic industries was reported in connection with this incident.
• Societal stability: Minor. The video aimed to cause panic and erode public trust during a crisis, but rapid action by platforms and the government minimized widespread societal destabilization.