Executive Summary
The October 2026 protests were born in France, and their causes lie in France. We found no evidence that a foreign actor was behind the origin of these protests. We did find, however, that foreign actors worked actively to amplify and inflame public anger, both before and after the protests, and to applaud violence, with the further aim of polarising the debates and dividing French society.
Inside France, foreign actors worked to fan the flames. Abroad, Russian state outlets, anglophone far-right influencers and US media reframed the unrest around migration and presented it as spreading across Europe.
Alliance4Europe’s study covers 547,471 posts published on X, Telegram, TikTok, YouTube, Facebook and news sites, in nine languages, as well as 12,839 comments posted under the most-viewed videos. We traced the Russian operations back to 1 September 2026, mapped the accounts running them and reconstructed their infrastructure: twenty-three of the network’s redirect sites share a single IP address.
In the six weeks before 17 October, a bot network that we attribute with high confidence to Russia’s Doppelganger operation posted 639 coordinated messages in French, German, English and Polish, seen roughly 7.4 million times. It seized on real grievances and turned them, dishonestly, against Ukraine: “Every euro for Kyiv is a euro stolen from our common prosperity” (translated from French). It invoked the memory of the 2018 “yellow vest” protests against fuel price rises and the carbon tax, mainly targeted centrist, democratic-left, pro-European and pro-Ukraine politicians, and praised both certain measures proposed by Jean-Luc Mélenchon and an unnamed “patriot” candidate. The operation is pragmatic and opportunistic: it backs no particular party, but pursues an objective that serves Vladimir Putin’s regime, namely that France stop supporting Ukraine.
Once the riots broke out, their story was picked up and distorted abroad. Roughly one post in thirty-three in French directly blamed migrants or Islam; in English, nearly one in five. The unfounded claims that immigrants had set fire to Limoges town hall circulated 3.6 times more in English than in French, and appeared in our data about six hours earlier in English. The Russian Pravda network published 1,018 articles across twenty-eight sites in nine languages, copying content from one site to the next within minutes. This content targets machines as much as human readers. A chatbot has already cited the network’s French site while repeating Kremlin falsehoods, and ten AI tools previously tested repeated the network’s claims in a third of cases. Researchers argue the network is also designed to reach the data used to train AI models, a tactic they call “LLM grooming”.
This is not only about France, nor even about Europe. In Canada, Russian operations and pro-Trump voices amplified the same separatist cause. Pages based outside Europe sell “decline” narratives to European audiences to generate clicks.
