Follow us on Google News
Leaked documents have revealed how John Mark Dougan, a former US Marine and law enforcement officer living in Russia, allegedly used artificial intelligence tools, including Anthropic’s Claude and Alibaba’s Qwen, to produce large volumes of pro-Russian content aimed at American audiences.
The material, reportedly obtained from Dougan’s laptop, includes emails, private messages and AI prompts detailing efforts to create websites and articles that appeared credible to Western readers. The operation allegedly sought to influence search engine results and increase the likelihood that AI chatbots would encounter and reproduce narratives favourable to Russia.
One leaked prompt attributed to Dougan claimed that the operation was producing approximately 43,000 articles every month at an estimated monthly cost of two million Russian rubles, or around $23,000.
Dougan, who also uses the online alias BadVolf, acknowledged that his material had been hacked and leaked.
FCRF’s Chief AI Officer Certification Returns on November 14 to Prepare Future AI Leaders
How Did the Alleged AI Propaganda Operation Work?
According to the leaked documents, Dougan used AI-powered content generation to produce and update thousands of articles promoting pro-Russian narratives.
The strategy reportedly involved creating websites that resembled legitimate news outlets and filling them with content designed to attract American readers.
The documents described efforts to improve the articles’ search rankings, readability and visibility to large language models.
By regularly publishing and updating AI-generated material, the operation allegedly sought to increase the likelihood that other AI systems would encounter the content when retrieving information.
The leaked records also contained details about the operation’s reported production costs.
In one prompt submitted to Claude, Dougan stated that the operation was writing 43,000 articles a month and sought assistance preparing a list of expenses.
The stated monthly expenditure was approximately two million rubles, equivalent to around $23,000.
These figures came from the leaked material and represent claims about the scale and cost of the operation.
The documents became public after Dougan’s emails, Telegram and Signal messages, and AI interactions were posted on a website resembling his own. A link to the material was also shared through his compromised X account.
How Were Claude and Qwen Used to Generate Content?
The leaked records identified Anthropic’s Claude and Alibaba’s Qwen among the AI systems used by Dougan.
They also referred to Open WebUI, a platform that allows users to interact with AI models.
According to the investigation, Dougan used Claude for tasks connected to developing websites and producing content intended to appeal to Western audiences.
One interaction showed him asking Claude for suggestions for a website promoting free speech in Europe, specifically targeting French-speaking readers.
The chatbot reportedly discussed whether a proposed website name would be suitable for a French audience.
The documents also showed Dougan approaching Alibaba’s Qwen for assistance with producing text that would be less likely to be identified by AI-content detection systems.
Such activities raise questions about the potential misuse of commercially available AI tools to support coordinated information campaigns.
Anthropic’s usage policies prohibit the creation of misleading information, fabricated identities and falsely attributed content. Alibaba also restricts the use of Qwen for misinformation.
The leaked exchanges indicate attempts to use these systems for activities that conflict with those restrictions.
However, the documents do not establish that Dougan compromised the underlying AI models or successfully altered their training data.
Who Is John Mark Dougan and What Do the Leaks Reveal?
Dougan previously served in the US Marine Corps and worked in law enforcement, including as a deputy sheriff in Palm Beach County, Florida, and Windham, Maine.
He was later dismissed from law enforcement positions following misconduct-related allegations.
In 2016, he came under FBI investigation over allegations involving the disclosure of confidential police records.
Dougan subsequently travelled to Russia, where he received political asylum after being indicted on 21 felony charges in Florida.
He was granted Russian citizenship in 2023.
Previous reporting has linked Dougan to pro-Russian information operations and alleged connections with Russia’s military intelligence service, commonly known as the GRU.
He has also been associated with websites designed to resemble established American news organisations while publishing material promoting Russian narratives.
Following the leak, Dougan acknowledged that his accounts and information had been compromised. He also confirmed the authenticity of the leaked material to investigators examining the documents.
The disclosures provide details of how AI systems were allegedly used to support the production and distribution of political messaging across websites aimed at Western audiences.
They also illustrate how large volumes of AI-generated content can be used in attempts to influence online information sources, search visibility and the material encountered by other AI systems.
About the author — Suvedita Nath is a science student with a growing interest in cybercrime and digital safety. She writes on online activity, cyber threats, and technology-driven risks. Her work focuses on clarity, accuracy, and public awareness.
