AI Chatbots Cite Russian State Media on Ukraine War
Research by the Institute of Strategic Dialogue found that four major AI chatbots cited Russian state media and propaganda sources in nearly one-fifth of responses to questions about the Ukraine war, raising concerns about their compliance with EU sanctions on Russian media.
What happened
According to research by the Institute of Strategic Dialogue, OpenAI's ChatGPT, Google's Gemini, DeepSeek, and xAI's Grok cited Russian state-attributed sources in nearly one-fifth of responses to questions about Russia's war in Ukraine.
The researchers tested the chatbots across multiple languages (English, Spanish, French, German, and Italian) using 300 neutral, biased, and "malicious" questions relating to NATO, peace talks, Ukraine's military recruitment, Ukrainian refugees, and war crimes, in an experiment in July 2025.
The chatbots cited Sputnik Globe, Sputnik China, RT, EADaily, the Strategic Culture Foundation, and the R-FBI, as well as Russian disinformation networks and journalists amplifying Kremlin narratives.
Reported impact
The ISD claims that Russian propaganda has exploited data voids where searches for real-time information provide few results from legitimate sources.
The findings raise questions about the ability of large language models to restrict sanctioned media in the EU, particularly as millions use chatbots as alternatives to search engines for real-time information.
Index analysis
Questions about peace talks between Russia and Ukraine led to more citations of Russian state-attributed sources than questions about Ukrainian refugees.
ChatGPT cited the most Russian sources and was most influenced by biased queries compared to the other three chatbots tested.
Governance controls
OpenAI stated that the company takes steps to prevent ChatGPT being used to spread false or misleading information linked to state-backed actors, and that these are long-standing issues the company is attempting to address.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.