Russian state actors are running a coordinated campaign to rewrite Wikipedia, and it's already poisoning AI models. Three separate research organizations have documented what they call the Pravda network: roughly 193 fraudulent news sites that launder pro-Kremlin propaganda into Wikipedia articles. French defense agency VIGINUM caught one case where a Pravda site was cited as a source for a Wikipedia page about a Red Sea naval operation, with edits traced to pro-Russian Telegram channels. VIGINUM calls this "foreign digital interference." That's exactly what it is. Russia isn't the only actor using AI to manipulate the information landscape. Here's why anyone building AI agents should care. Wikipedia is core training data for most large language models. When Kremlin-manipulated narratives get baked into Wikipedia, they get baked into the models too. The Institute for Strategic Dialogue used semantic clustering to detect coordinated manipulation across the English-language Wikipedia entry for the Russo-Ukrainian war and 48 related pages. Author Heidi Siegmund Cuda describes finding a Wikipedia page for a book about Soviet abuses that had been rewritten into a "sanitized fairytale where Putin is good." This is a sustained, well-resourced information operation that's been running since at least 2022. Major AI labs have defenses. They filter pre-training data to prioritize stable, high-trust Wikipedia versions and downweight edits from low-reputation accounts. They use reinforcement learning from human feedback and Constitutional AI to align models against biased outputs. Red teaming operations probe models for disinformation. But these tools catch obvious spam. Subtle, high-quality disinformation slips through. The Atlantic Council's Digital Forensic Research Lab confirmed direct Russian involvement with the Pravda network, which targets more than 80 countries and regions. If state actors can systematically distort the information ecosystem that AI systems learn from, we're all working with corrupted inputs. JD Vance and Marjorie Taylor Greene recently promoted a fake story traced to a Russian disinformation network about yachts purchased with Ukraine military aid. The same machinery rewriting Wikipedia is feeding social media, and it's working. State propaganda has become more sophisticated with the rise of generative AI.
Russia's Pravda Network Rewrites Wikipedia, Poisons AI
Russian state actors are running a coordinated campaign to rewrite Wikipedia through 193 fraudulent news sites, and the manipulated narratives are already poisoning AI training data. Research from VIGINUM, the Institute for Strategic Dialogue, and the Atlantic Council documents how the Pravda network launders pro-Kremlin propaganda into Wikipedia and LLMs.