AI Chatbots Caught Sending 300,000 Users to Russian Propaganda Sites

AI Becomes Unexpected Gateway to Kremlin Content

New web analytics data has exposed a troubling trend: major AI chatbots directed approximately 300,000 visits to sanctioned Russian state media and pro-Kremlin propaganda websites during the final quarter of 2025.

The findings, based on SimilarWeb data for October through December, show that conversational AI platforms including ChatGPT, Perplexity, Claude, and Mistral sent users to at least eight Kremlin-linked websites, including RT and Sputnik domains.

ChatGPT Emerges as Largest Referrer

ChatGPT emerged as the largest identifiable referrer, sending more than 88,000 visits to RT alone during the quarter. Similar patterns appeared across other Russian-language outlets, with RIA Novosti receiving more than 70,000 AI-driven visits and Lenta.ru recording over 60,000.

For smaller propaganda outlets, the AI contribution proved even more consequential. On Sputnik Globe, ChatGPT and Claude together represented roughly six percent of all referral traffic, while Claude-generated traffic accounted for nearly ten percent of referrals to News Pravda, a multilingual disinformation website known for deliberately targeting AI systems.

The Data Void Strategy

Researchers say the pattern exploits what they call "data voids", areas where legitimate information is scarce, allowing false narratives to fill the gaps. A Russian disinformation network called Pravda has published more than 3.6 million articles specifically designed to seed disinformation into the datasets AI models use.

A study by the Institute for Strategic Dialogue found that nearly eighteen percent of chatbot responses about the Ukraine war cited sources attributed to Russian state media. Separately, NewsGuard reported that leading chatbots repeated false narratives pushed by the Pravda network about one-third of the time.

Regulatory Response Intensifies

The European Commission updated its sanctions guidance on 22 January 2026 to explicitly prohibit AI services to Russia, including access to large language models. However, enforcement remains challenging as chatbots function as intermediaries rather than direct distributors of sanctioned content.

OpenAI has stated it is working to prevent ChatGPT from spreading content associated with state-backed entities, while the European Commission maintains it is the responsibility of platform providers to block access to sanctioned outlets.