This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A Russian disinformation network named Storm-1679 or Matryoshka deployed AI-generated deepfake videos and false news reports impersonating major media outlets to influence US elections and public opinion.
A Russia-linked Matryoshka disinformation operation reportedly circulated a deepfake video falsely branded as Mother Jones that impersonated editor Clara Jeffery and claimed the DNC paid CNN chief Mark Thompson $259 million to promote gender transition among children. Two X posts displayed about 205,000 views, though researchers said impressions may have overstated authentic reach. X later removed the posts after Mother Jones complained.
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 campaign generated and spread highly inflammatory, derogatory, and hateful synthetic statements falsely attributed to celebrities and political figures.
The operation systematically forged the likenesses, voices, and branding of professional journalists, media organizations, and creative artists, exploiting and devaluing their public reputations.
The incident involves a coordinated, state-aligned foreign influence network using AI-generated deepfakes and synthetic media to conduct large-scale disinformation campaigns aimed at manipulating US political processes and public opinion.
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 risk is caused by human malicious actors intentionally deploying AI-generated deepfakes and synthetic media post-deployment to conduct influence operations.
Added categories peak at 4: Severe (Epistemic Harm, Inferred)
A Russian state-aligned disinformation network utilized AI deepfakes and cloned voices to impersonate major media outlets and celebrities, successfully injecting false narratives into the feeds of millions of Americans to influence US elections. This represents a substantial information warfare and sovereignty threat, though it lacks physical or kinetic dimensions.
The AI systems (deepfake generators and voice cloning software) were used as simple tools by human malicious actors to generate synthetic video and audio content based on direct inputs, without exhibiting independent agency or autonomous planning.
Limited Risk: The incident involves the generation and dissemination of AI-generated content, specifically deepfakes, which are subject to transparency and disclosure obligations under the EU AI Act.
The primary technology used in the incident was the generation of highly realistic synthetic videos and cloned voices of celebrities and journalists to deceive the public.
The report provides clear, detailed, and consistent accounts of a coordinated foreign disinformation campaign utilizing synthetic media, citing specific fake videos, view counts, targeted individuals, and the research groups tracking them.
None; the report clearly frames this as a deliberate, politically motivated foreign influence operation using synthetic media.
While the specific technical tools and models used by the Russian network are not identified, this does not prevent confident classification of the incident's domain, causes, and harms.
Technology
Generative AI Video Synthesis
Creation of highly realistic deepfake videos mimicking celebrities and media.
Algorithmic Amplification
Social media recommendation algorithms elevated the reach of fake posts.
Automated Bot Networks
Use of automated bot accounts to quickly spread and inflate view counts.
Data Inputs
Unauthorized Trademark Use
Misuse of media logos like CNN and Mother Jones to establish false trust.
Stolen Personal Emails
Use of pilfered personal emails to provide false context for disinformation.
Doctored Media Footage
Manipulation of authentic video footage of public figures to change context.
Human Factors
Influencer Amplification
High-profile individuals reposted deepfakes to millions of followers.
Targeted Political Biases
Exploitation of partisan divides to make the fake stories believable.
Lack of User Verification
Social media users reshared sensational content without verifying sources.
Process and Methods
Reactive Moderation Policies
Platforms only removed content after receiving formal trademark complaints.
Inadequate Detection Systems
Social media platforms lacked automated systems to block obvious deepfakes.
Delayed Incident Reporting
Dependence on external activist groups to identify and flag coordinated campaigns.
Management
Prioritizing Engagement Over Safety
Platform management allowed unverified viral content to spread unchecked.
Insufficient Safety Investment
Inadequate resource allocation to preemptively detect coordinated bot networks.
Regulatory Environment
Dismantled Government Oversight
Closure of federal units tracking and countering foreign disinformation.
Lack of Platform Accountability
Absence of strict regulations forcing platforms to proactively scan for 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.
