This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
John Mark Dougan, a U.S. fugitive residing in Russia, orchestrated a large-scale disinformation network consisting of over 160 fake news websites. These sites utilized AI tools like ChatGPT and DALL-E 3 to mass-produce articles, images, and fake journalist personas to spread Russian propaganda and influence U.S. and European elections. The operation has been linked to Russian military intelligence (GRU) and has successfully reached millions of viewers, with some false narratives being amplified by mainstream political figures.
John Mark Dougan, a former Florida sheriff's deputy granted asylum in Russia, has been implicated in spreading disinformation. Utilizing AI tools like OpenAI's ChatGPT and DALL-E 3, Dougan created over 160 fake news sites, disseminating false narratives to millions worldwide. His actions align with Russian disinformation strategies targeting Western democracies. See also Incident 734.
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 network generated and spread highly targeted, defamatory, and malicious deepfakes and articles, such as false allegations of sexual misconduct targeting political candidates.
The network of over 160 fake news sites flooded the information ecosystem with AI-generated articles, which were then ingested and repeated by major commercial chatbots.
The incident describes a coordinated, state-linked influence operation using AI to generate and scale disinformation targeting political processes and public opinion across multiple countries.
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 (John Mark Dougan and Russian state-sponsored operatives) intentionally using deployed commercial AI models to generate and scale disinformation.
Added categories peak at 4: Severe (Epistemic Harm, Inferred)
A massive, Russian GRU-linked disinformation network led by a US fugitive utilized commercial generative AI tools to scale influence operations targeting US and European elections. By generating deepfakes and automating articles across 160 fake news sites, the operation successfully spread propaganda to millions, demonstrating how AI lowers the cost and increases the reach of state-sponsored information warfare.
The AI systems (ChatGPT, DALL-E 3) were used as tools by human operators to perform specific, well-defined functions such as generating text and images based on direct prompts, without exhibiting independent planning or autonomous decision-making.
Limited Risk: The incident involves the use of AI systems to generate text and deepfakes, which are subject to transparency obligations under the EU AI Act to ensure users are aware they are interacting with AI-generated content.
The primary goal of using the AI models was to generate high-volume, persuasive text and media content to populate fake news websites and social media feeds for influence operations.
The reports provide detailed, consistent, and well-sourced accounts from multiple independent organizations (NewsGuard, BBC Verify, Microsoft, Clemson University) detailing the operation, the specific AI tools used, and the reach of the disinformation campaigns.
None
Technology
Generative AI Content Scaling
Commercial AI tools enabled rapid scaling of low-quality propaganda articles.
AI Voice Cloning and Deepfakes
AI cloned the FBI director's voice and created fake videos to spread fraud claims.
Algorithmic Prioritization of Engagement
Social media algorithms prioritized high-engagement extreme content over truth.
Data Inputs
Chatbot Training on Propaganda Sites
Chatbots ingested fake local news sites, repeating Russian narratives 32 percent of the time.
Lack of Source Reliability Filtering
AI models failed to recognize propaganda front sites like 'The Boston Times'.
Plagiarized and Rewritten Content
Generative AI was used to rewrite plagiarized articles from mainstream media.
Human Factors
User Susceptibility to Sensationalism
Audiences easily shared shocking deepfakes and stories without verification.
Influencer Amplification of Falsehoods
High-profile influencers shared fake narratives, exposing them to millions.
Trust in Chatbots for Accurate Info
Users increasingly relied on AI chatbots for election-related information.
Process and Methods
Narrative Laundering Techniques
Using paid actors and fake sites to make disinformation appear as legitimate news.
Inadequate Chatbot Guardrails
AI models lacked robust filters to prevent repeating debunked conspiracy theories.
Evasion of Platform Moderation
Operations used redirect chains and cloaking to bypass social media detection.
Management
Inadequate Platform Verification
Tech platforms failed to proactively prevent mass creation of fake accounts.
Delayed Action on Fake Domains
Adversaries easily recreated seized domains within 24 hours under new TLDs.
Prioritizing Outreach Over Safety
AI companies deployed models without ensuring immunity to foreign propaganda.
Regulatory Environment
Lack of Global AI Safety Standards
Absence of enforceable international regulations on generative AI deployment.
Permissive Hosting Environments
Bulletproof hosting providers allowed malicious sites to remain online.
Weak Know-Your-Customer Requirements
Hosting services permitted purchasing infrastructure without identity verification.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
