This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Greek economist and politician Yanis Varoufakis reports that hundreds of deepfake videos using his face and voice have been circulating on YouTube and social media. These videos depict him making fabricated political statements, leading to public confusion and reputational harm. Despite attempts to have the content removed by platforms, the videos continue to reappear, highlighting the challenges of managing synthetic media impersonation.
Purported deepfake videos using the face and voice of Greek economist and politician Yanis Varoufakis were reportedly circulated on YouTube and other social platforms in late 2025 and early 2026. The videos depicted a synthetic version of Varoufakis delivering fabricated political statements, including about international events. Platforms reportedly removed some videos, but many reappeared under new accounts.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to gain a personal advantage over others such as through cheating, fraud, scams, blackmail or targeted manipulation of beliefs or behavior. Examples include AI-facilitated plagiarism for research or education, impersonating a trusted or fake individual for illegitimate financial benefit, or creating humiliating or sexual imagery.
Also classified as
The deepfakes generate false political statements, such as fictitious claims about a coup in Venezuela, misleading the public.
The deepfakes have proliferated at scale across major platforms to muddy public political discourse and drown genuine dissent.
The incident involves the unauthorized use of a public figure's likeness to create deepfakes for targeted manipulation, reputation damage, and political impersonation.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The incident was caused by malicious human actors intentionally creating and distributing deepfake videos of the victim post-deployment.
Added categories peak at 4: Severe (Epistemic Harm, Inferred)
A persistent deepfake campaign targeted Greek politician Yanis Varoufakis with fabricated political statements, causing public confusion and reputational harm. While demonstrating the evolving challenge of synthetic media and platform moderation, the incident has minor national security implications and did not disrupt critical sovereign functions.
The AI system was used as a generative tool to synthesize the victim's face and voice based on direct inputs. The report describes the technology being used to produce specific synthetic media outputs without any autonomous planning or interactive assistant behavior.
Limited Risk: The report describes the creation and proliferation of 'AI-generated content such as deepfakes', which falls under the transparency obligations of Limited Risk systems.
The AI systems were used to generate highly realistic synthetic videos ('deepfake AI doppelganger') and synthesize the victim's voice ('synthesising my voice') to impersonate him.
The report is a first-hand account by the victim, providing clear details about the nature of the deepfakes, the platforms used, the impact on his reputation, and his attempts to have them removed. While the specific AI models are not named, the description of the synthetic media and its effects is highly detailed and unambiguous.
None
Technology
AI Voice and Face Synthesis
Creates highly persuasive synthetic videos of individuals without their consent.
Algorithmic Recommendation Feeds
Amplifies deepfakes by prioritizing engagement over content authenticity.
Data Inputs
Publicly Available Media Data
Original videos are harvested to train and clone the victim's voice and face.
Human Factors
Audience Confusion
Viewers struggle to distinguish between real and deepfaked political content.
Weaponization by Opponents
Opponents circulate synthetic videos to damage the victim's reputation.
Process and Methods
Slow Platform Takedown Processes
Reporting processes take a week or more to remove infringing deepfake videos.
Ineffective Reappearance Barriers
Banned channels instantly recreate profiles to re-upload deepfake content.
Management
Prioritizing Engagement Over Safety
Platforms focus on cloud rents and traffic rather than proactive verification.
Absence of Proactive Verification
Platforms lack automated systems to detect and block deepfakes before upload.
Regulatory Environment
Lack of Digital Identity Protections
No legal framework prevents the unauthorized AI replication of human identity.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
