A new report by the Institute for Strategic Dialogue (ISD) has revealed that major AI chatbots — ChatGPT, Grok, DeepSeek and Gemini — are inadvertently spreading content from Russian state media banned in the European Union. According to the investigation, when asked about topics such as the war in Ukraine or NATO, these chatbots occasionally cite material from outlets including Sputnik, RT and EADaily, which have been sanctioned for disseminating Kremlin propaganda. The report warns that loopholes in search engine datasets allow Russian disinformation to bypass EU restrictions through AI-generated responses.
Disinformation through data gaps
ISD researchers tested the four chatbots in July and October 2025, sending 60 multilingual queries covering sensitive geopolitical issues — from peace talks and refugee crises to Russian war crimes in Ukraine. The study found that 18% of responses contained links to Russian or pro-Kremlin sources. ChatGPT was the most likely to cite Russian outlets, while Grok frequently referenced pro-Russian bloggers, and Gemini provided the most accurate answers with clear source labeling. The report attributes these distortions to “data gaps,” where AI systems, lacking sufficient verified material, pull content from untrustworthy or sanctioned domains.
Propaganda disguised as alternative views
The study also noted that some blogs and websites appearing neutral were, in fact, amplifying Moscow’s narratives under the guise of “alternative perspectives.” Neutral questions triggered references to propaganda sources in 11% of cases, but that rate doubled when users submitted biased or evidence-demanding prompts. ISD analysts warned that this pattern mirrors the Kremlin’s own manipulation tactics, exploiting user phrasing to push favorable content into mainstream discourse through AI systems.
Risks for EU information security
Researchers caution that as millions of Europeans turn to AI chatbots for news and fact-checking, even a small share of manipulated responses could erode trust in the EU’s information ecosystem. The spread of such “toxic narratives” risks normalizing anti-Ukrainian and anti-European rhetoric, undermining public understanding of the war and weakening resilience against disinformation campaigns. The report frames this as an emerging threat to the bloc’s “informational sovereignty.”
Calls for transparency and regulation
ISD recommends combining technical safeguards with regulatory oversight. It proposes that chatbot providers implement mandatory filters synchronized with EU sanctions lists and conduct independent audits of how their models handle sensitive topics such as war and elections. The report urges platforms to clearly disclose their information sources and explain why particular links appear in their responses. Without such measures, experts warn, artificial intelligence could become a major enabler of Russian propaganda within Europe.
