Russians Manipulate AI Chatbots. Poland as a Target.

“A staggering 3,600,000 Russian propaganda articles published in 2024 — are now being incorporated into the outputs of Western AI systems, contaminating their responses with false claims and propaganda” wrote NewsGuard researchers McKenzie Sadeghi and Isis Blachez in a report.

NewsGuard’s audit of the ten most popular generative AI tools, including OpenAI’s ChatGPT-4o, Google’s Gemini, and Microsoft;s Copilot, revealed a disturbing pattern: in 33% of test cases, these models repeated false narratives sourced from Pravda. This indicates a significant shift in the tactics employed to spread disinformation: rather than directly targeting audiences, Moscow;s information
warfare apparatus is poisoning the data streams that AI models rely on to generate responses.

“The Pravda networks ability to spread disinformation on such scale is unprecedented, and its potential to influence AI systems makes the threat even more dangerous” says Nina Jankowicz, Executive Director of the American Sunlight Project, as quoted by AFP.

The Russian Pravda Network: How the Biggest Pro-Kremlin Disinformation Machine Operates Today

1. Origins and Structure
Established in Russia during the spring of 2022, the Pravda disinformation network was created by the same team which was behind the “Doppelgänger”; operation, known for cloning Western media websites. It centres around hundreds of domains including news-pravda.com , pravda-fr.com , pravda-pl.com , etc., each designed to mimic the appearance of local news services.

2. Scale and Automation
The network has published over 3.7 million articles in 50 languages—most in France (394,000), Germany (377,000), Ukraine (270,000), and Poland. Real-time data on the network’s activities can be viewed at: Pravda Dashboard, https://solatrix.github.io/Pravda-Dashboard /.

3. Content is almost entirely generated through automated processes, sourced from TASS, RIA, Lenta, RT, and other Russian media outlets. In 2025, the network began generating clickbait headlines and leads using LLMs (Large Language Models), flooding search engines with low-quality content. These advanced AI models are trained on vast amounts of text data and are capable of processing and generating natural-sounding language.

4. It is important to recognise that the Pravda Network doesn’t necessarily aim toconvert everyone to the Russian narrative; but it seeks to drown public debate in noise, thereby reducing trust in facts. An effective defence is to swiftly expose these tactics and build public resilience to low-quality content.
According to analysts from the Atlantic Council and DFRLab, the Pravda network produces millions of manipulated and false pieces annually, which are often translated into multiple languages. These include fake quotes, doctored images, and manipulated headlines that appear on websites styled as local media in Europe, Asia, or Latin America.
In countries such as Germany, France, Poland, and the U.S., users may unknowingly land on what appears to be a local news portal—when in fact it is operated from Moscow or via intermediaries, for example, entities registered in the Balkans.

Poland as a target

In Poland, the Pravda network aggressively promotes narratives such as the “Ukrainization” of the country, cultural threats posed by refugees, the alleged involvement of Polish troops in Ukraine, imperial ambitions toward western Ukraine, and the social and economic toll of supporting Ukraine.
Importantly, these narratives are often accompanied by authentic images, statements taken out of context, and fabricated quotations attributed to real politicians, journalists, or experts.

Mechanisms of Operation

The Pravda Network uses a full arsenal of digital information warfare tactics:
1. Bots and troll farms—automated accounts and paid commentators that flood
forums and social media;
2. Mirror and fake mirror sites—websites designed to look identical to well-known media outlets (e.g., BBC, “Le Monde”) that publish manipulated content;
3. Redistribution networks—false texts published across multiple sites to create the illusion of mass media coverage;
4. Smuggled narratives—content disguised as analysis that subtly introduces pro-
Russian messaging.

Written by ih
Sources:

A well-funded Moscow-based global ‘news’ network has infected Western artificial intelligence tools worldwide with Russian propaganda

Russian disinformation ‘infects’ AI chatbots, researchers warn

A Well-funded Moscow-based Global ‘News’ Network has Infected Western Artificial Intelligence Tools Worldwide with Russian Propaganda