This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Russian state television broadcast a deepfake video featuring AI-generated audio of Ukrainian security official Oleksiy Danilov, falsely suggesting Ukrainian involvement in the Crocus City Hall terrorist attack. The video was identified as a composite of existing interviews with AI-generated audio dubbed over the footage. This incident is described as part of a disinformation campaign to influence public opinion and justify further aggression against Ukraine.
Russian state television reportedly aired a fabricated video that appeared to imitate senior Ukrainian security officials and present a false admission of involvement in the Crocus City Hall massacre. The clip reportedly used elements from recent interviews combined with allegedly AI-generated audio to create a misleading confession. ISIS-K is reported to have publicly claimed responsibility, while independent analysts and governments have rejected the narrative promoted through the video.
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.
Also classified as
The AI-generated video and audio fabricated false statements from a public official to mislead the public about a major terrorist event.
The deepfake was broadcast on Russian state television as part of a coordinated disinformation campaign aimed at manipulating public opinion and influencing geopolitical dynamics.
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 created and broadcasted a deepfake video using deployed AI tools to spread political disinformation.
Added categories peak at 4: Severe (Epistemic Harm, Direct)
Russian state television broadcasted a deepfake of Ukrainian security official Oleksiy Danilov falsely claiming responsibility for the Crocus City Hall terrorist attack. The incident represents a substantial information warfare operation aimed at manipulating public opinion and justifying further military aggression against Ukraine, highlighted by rapid international efforts to debunk the fabrication.
The AI system was used as a tool to generate synthetic audio and video (deepfake) based on direct inputs and instructions from human creators, without exhibiting autonomous planning or agency.
Limited Risk: The report describes an AI-generated deepfake video, which falls under the transparency obligations of Risk Level 3.
The AI system was used to generate a deepfake video of Oleksiy Danilov with synthetic audio dubbed over existing footage.
The reports from multiple reputable sources consistently describe the broadcast of the deepfake, its content, and how it was debunked by experts like BBC Verify.
None. The consensus is clear that the video was a fabricated deepfake.
Technology
AI Voice Cloning
AI cloned Danilov's voice using audio from recent interviews.
Video Mashup Software
Software allowed mashing of two different video interviews into one.
Data Inputs
Source Video Availability
Recent high-quality TV interviews provided the source data for the deepfake.
Human Factors
Disinformation Intent
Malicious actors created fake content to manipulate public perception.
Public Susceptibility
Indoctrinated audiences are highly susceptible to believable deepfakes.
Process and Methods
No Authentication Process
Broadcasters failed to verify the authenticity of the video before airing.
Rapid Content Assembly
Quick generation of deepfakes allows rapid response to real-world events.
Management
Prioritizing Narrative
Management prioritized political narratives over journalistic truth.
Regulatory Environment
Unregulated AI Media
No regulatory framework prevents state media from broadcasting deepfakes.
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.
