A new report from UK-based think tank Demos, shared exclusively with Euronews's fact-checking team The Cube, found that mainstream large language models (LLMs) such as Google Gemini have cited pro-Russian propaganda from an EU-sanctioned outlet. Researchers selected 50 propaganda narratives from ten articles published by the Foundation to Battle Injustice or "r-FBI", a fake human rights organisation founded in 2021 by deceased former Wagner Group leader Yevgeny Prigozhin and sanctioned by both the US and EU. The claims included that Ukraine is reusing the corpses of dead soldiers for meat products and that Zelenskyy drains the crypto grid. Testing five mainstream LLMs—operated by xAI, OpenAI, Anthropic, Google Gemini, and Mistral AI—in English, German, and French, researchers found all five cited r-FBI as a favourable source. A total of 16.6% of responses engaged with r-FBI's material in a way that could serve propaganda interests, 30.9% failed to surface the sanctioned-source context, and 52% rejected or debunked false claims. When questioned about allegations of processed human meat, OpenAI's GPT 4.1 Mini cited an r-FBI article as evidence.