This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A deepfake video of Bulgarian politician Ahmed Dogan was created by overlaying AI-generated audio onto authentic footage from an August 27, 2024, meeting. The video, which falsely depicted Dogan urging party members to protest to protect his wealth, was disseminated via a satirical TikTok account and subsequently amplified by anonymous Facebook profiles linked to political groups. This incident highlights the use of synthetic media to manipulate political discourse and spread disinformation.
A reportedly AI-manipulated video appeared to repurpose footage of Ahmed Dogan, honorary chairman of Bulgaria's Movement for Rights and Freedoms (DPS), by overlaying synthetic audio that made him seem to urge party members to protest in defense of his wealth. The clip reportedly originated from a satirical TikTok account and was later circulated on Facebook by anonymous profiles.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to conduct large-scale disinformation campaigns, malicious surveillance, or targeted and sophisticated automated censorship and propaganda, with the aim of manipulating political processes, public opinion, and behavior.
The deepfake video was used in a coordinated manner across social media platforms to manipulate political discourse and influence public opinion during an internal party conflict.
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 human actors who intentionally generated and distributed a deepfake video of a politician using deployed AI voice generation tools.
Added categories peak at 2: Minor (Epistemic Harm, Direct)
An AI-generated voice clone was used to create a deepfake of Bulgarian politician Ahmed Dogan, falsely depicting him calling for protests. While the incident represents a concerning use of synthetic media to manipulate political discourse during an internal party conflict, its national security impact remains minor as it was quickly debunked and did not cause widespread instability.
The AI system was used as a tool to perform the specific function of generating synthetic audio based on direct inputs, without autonomous planning or decision-making.
Risk Level 3. Limited Risk: The incident involves AI-generated content (a deepfake), which poses a moderate risk and is subject to transparency obligations under the EU AI Act.
The AI was primarily used to generate a synthetic voice clone of the politician, which was then used to create a deepfake video.
The fact-checking report provides clear and verified details about the creation, content, and dissemination of the deepfake video, as well as the political context surrounding it.
None. The video is clearly identified as a politically motivated deepfake.
Technology
AI Audio Generation
AI was used to generate realistic synthetic audio of the politician's voice.
Deepfake Video Overlay
Synthetic audio was overlaid on authentic video footage from a party meeting.
Data Inputs
Authentic Video Footage
Real footage from the DPS meeting on August 27 2024 was used as the base.
Human Factors
Malicious Dissemination
Anonymous profiles shared the deepfake to influence political DPS factions.
Satirical Creation
A satirical TikTok account initially created and shared the deepfake video.
Process and Methods
Anonymous Account Creation
Platforms allowed fake profiles with stock photos to share content.
Lack of Platform Filtering
TikTok and Facebook failed to detect and label the deepfake audio promptly.
AI system involved
An AI system was involved, but the reports do not identify it.
Identified by automated extraction; can differ from the alleged developers above.
