The far-right has developed a strategy to spread racist, antisemitic, and anti-immigration messages online using coded language. These terms, which appear neutral or even benign to the general public, allow them to bypass content moderation algorithms and avoid legal consequences. By using specific phrases or words, they can communicate with like-minded individuals while staying under the radar. This method is known as a « dog whistle, » a term that describes a message that only those who understand it can fully grasp. « It's a signal sent to those who know it, » explains Daphné Deschamps, a journalist at StreetPress. Examples of such coded language include « Noix, » used to refer to Black people, « Arbres, » for Arabs, and « Qui ? » for Jews. These terms are often used in the context of online discussions to subtly express prejudiced views without being overt. The use of coded language presents a legal challenge. While the context of these terms can make them difficult to interpret, a judge may still find them punishable if they clearly reference radical ideas, warns Baptiste Buissart, a criminal lawyer and cybercrime expert at the Catholic University of Lille. This coded communication is not only effective but also strategic. « When you speak in dog whistle, the evocative power is very powerful, it's as if you were caressing other parts of the brain, » says Illana Weizman, a sociologist and author. One example of this is the phrase « It's Nicolas who pays, » originally used by a streamer to contrast the archetype of a white Frenchman with immigrants and retirees. The phrase later became popular online and is now used to imply that white people are the ones bearing the cost of immigration. Social media algorithms have evolved over time, especially after the pandemic and with changes in platform leadership, such as Elon Musk's takeover of X. These changes have led to more flexible content moderation policies, allowing previously banned individuals to return and spread their ideas. For instance, some personalities banned from YouTube have found ways to reappear by being invited to share their views, such as Alain Soral. Even if these individuals are « shadowbanned »—meaning their content is de-prioritized by algorithms—their popularity can still make these sanctions ineffective. « They have such notoriety that people will go look for them on their own, » notes Romain Fargier, a researcher specializing in digital culture. Other tactics include using live streams to avoid detection. Streamers may broadcast for long periods with a lot of content, making it harder for platforms to identify offensive language within the stream. Some influencers also use less radical content to attract a wide audience and then redirect them to private forums where more extreme views can be shared. As the 2027 French presidential election approaches, and the far-right is seen as a potential favorite, these coded messages and strategies could become even more prevalent online. The use of terms like « Noix » and « Arbres » may increase, further complicating efforts to monitor and regulate harmful content.