PARIS — In early August 2026, a fabricated video testimony attributed to the veteran investigative journalist Edwy Plenel began circulating online, appearing to confirm that his co-host, Léa Salamé, had paid major news outlets to secure favourable coverage for her partner, the presidential candidate Raphaël Glucksmann. The clip was quickly exposed as an AI-generated deepfake, and France’s national security services later traced it to Storm-1516, a Russian influence network run by the GRU. The episode was not an isolated piece of election mischief. It was a demonstration of a broader strategy that French and allied intelligence agencies say is designed to shape the 2027 French presidential election through cheap, redundant and deliberately hard-to-neutralize operations. These campaigns do not need to persuade voters to believe any single lie. Their purpose is to pollute the information environment so thoroughly that trust in candidates, institutions and the media is eroded before the first ballot is cast. Debunking a specific fabrication, officials and researchers warn, cannot erase the residue of confusion and suspicion that the lie leaves behind. The Glucksmann-Salamé deepfake was one of three known operations targeting presidential candidates within weeks. Édouard Philippe, a former prime minister, was subjected to fabricated rumours that he had frontotemporal dementia. Gabriel Attal, another former prime minister, was falsely accused of having Parkinson’s disease. In each case, the alleged evidence was manufactured, the distribution followed the same pattern of burner accounts and amplifier websites, and the intended effect was not to win a debate but to cast doubt on the integrity and fitness of the candidates. The choice of targets was not random. Philippe has been a vocal supporter of continued European military backing for Ukraine. Attal served as prime minister under a staunchly pro-Ukrainian president. Glucksmann has built his political career on European federalism and support for Kyiv. None of the three is aligned with Marine Le Pen, the leading candidate in first-round polling and the major contender most sympathetic to Moscow. Le Pen remains eligible to run despite her conviction over the misuse of European Parliament funds. French authorities, including VIGINUM, the agency tasked with countering foreign digital manipulation, have framed the strategic objective of these networks as the erosion of Western support for Ukraine.
Russia has spent years cultivating ties with the French far right and working to erode public support for Ukraine, and the summer of 2026 offered a preview of how those efforts translate into a presidential campaign. Between July and August, three candidates were targeted by some form of Russian network activity, according to French officials. Before the Glucksmann-Salamé deepfake, Philippe and Attal were each targeted by fabricated rumours about their health. The Philippe narrative claimed he had frontotemporal dementia, a degenerative neurological condition; the Attal narrative accused him of having Parkinson’s disease. Both were false, but both were engineered to look like leaked medical information, giving them an air of authenticity that made them difficult to dismiss without an official response. The pattern is significant because of what it reveals about the strategic logic of the operations. All three men have been outspoken in their support for Ukraine. Philippe has repeatedly called for sustained European military aid, Attal served in the government of a president who made support for Kyiv a central pillar of foreign policy, and Glucksmann has positioned himself as one of the most pro-European and pro-Ukrainian figures in French politics. By contrast, Le Pen, who has long maintained sympathetic positions toward Moscow, was not targeted. That asymmetry is not accidental. Analysts at VIGINUM and allied agencies have concluded that the objective is not to promote Le Pen directly but to weaken the credibility of the candidates most associated with the transatlantic and European consensus on Ukraine. In a campaign where foreign policy may be secondary to domestic concerns, the goal is to make the pro-Ukraine camp look compromised, unstable or untrustworthy before voters even begin to weigh their options.
The mechanics of these operations have been mapped in detail over the past two years. In a May 2025 report, VIGINUM documented 77 Storm-1516 campaigns between the network’s emergence in August 2023 and March 2025. While the nature of the manufactured narratives varies, the playbook is consistent. Disposable social media accounts, often acquired in bulk from suppliers, insert fabricated evidence under the guise of “whistleblowing” or exposing alleged wrongdoings. These accounts are designed to be sacrificed once they attract attention; their only job is to seed the story. Paid third-party sites and other amplifiers then launder the content by republishing it as though it were independently sourced, giving it a veneer of legitimacy. Pro-Russian sympathizers, often unpaid and ideologically motivated, spread it further, no longer spurred on by monetary compensation. Microsoft’s Threat Analysis Center has found that the network’s strength lies in creating a supply of “outlandish fake videos and scandalous claims,” then using affiliated websites to give those claims wider reach. Clint Watts, the centre’s general manager, has described the network’s model as one of sheer volume rather than precision. Most fabrications fail to land, but at that scale, occasional success is close to guaranteed. The redundancy is built into the system: if one account is suspended, another takes its place; if one website is taken down, another appears; if one narrative is debunked, another is already in circulation. This is why the response to any single piece of disinformation is inherently limited. The network does not depend on any one lie succeeding, and it does not collapse when a lie is exposed.
The paradox of debunking is that it can inadvertently amplify the very rumours it seeks to correct. Major news outlets and mainstream media have more rigorous fact-checking processes, and they often shut down manufactured rumours before they reach a nationwide audience. But the act of debunking itself extends the reach of the original claim to a much larger audience. A voter who had never seen the Glucksmann-Salamé deepfake might first learn about it from a news article explaining that it was fake. At the same time, the correction publicizes the fact that the video existed and that it was taken seriously enough to warrant an official response. The specific content of the lie ends up meaning less than the fact that it existed in the first place. Researchers describe this as the “liar’s dividend”: in an environment saturated with synthetic media, authentic evidence starts to lose credibility too, since anything, real or fake, can be plausibly dismissed as fabricated. If that dynamic takes hold, the risk is not only that voters believe specific lies about specific candidates; it is that they stop trusting verification itself, which would make any future correction, however accurate, less persuasive by default. Glucksmann himself described the goal of these operations as creating a “halo of suspicion,” in line with the logic that where there is smoke, there’s fire. Exposure and correction cannot fully erase the suspicion that a fabricated claim leaves behind. Multiplied across dozens of operations run by several independent networks, this phenomenon becomes nearly impossible to contain through debunking alone. Catching one burner account, one fake site, or even one entire network does nothing to another network already active elsewhere. Three candidates targeted in a matter of weeks, by two independent operations, is a preview of what that redundancy looks like at scale over a full campaign.
France’s institutional response has moved quickly, at least on paper. Prime Minister Sébastien Lecornu credited a new detection system for automatically flagging all three campaigns as they emerged, and said a draft law addressing digital interference had already been presented to the Council of Ministers. The proposed legislation is expected to give French authorities new powers to identify, attribute and sanction foreign disinformation operations, including the ability to demand the removal of content and to impose penalties on platforms that fail to act. Separately, the Paris prosecutor’s office opened formal investigations into the operations targeting Glucksmann, Attal and Philippe. These investigations are significant because they treat the campaigns not merely as propaganda but as potential criminal interference with the electoral process. Whether faster detection and a new legal framework can outpace a pipeline built specifically to survive being caught is, at this stage, an open question rather than a settled outcome. The networks have already demonstrated an ability to adapt to takedowns and to shift tactics when their methods are exposed. Yet the French response marks the first sign that institutions are treating the redundancy of the system, rather than each individual fabrication, as the problem to solve. The focus is no longer solely on debunking false stories after they appear; it is on disrupting the infrastructure that makes them possible, from the suppliers of burner accounts to the platforms that host the amplifier sites.
The central challenge is not simply stopping false stories before they spread. These campaigns do not need to convince voters of any particular lie to be effective. Their success can instead be measured by how much uncertainty they create. Enough contradictory claims, fabricated evidence and competing narratives can make voters less certain about what is true and less willing to trust the institutions responsible for telling them the truth. In that sense, the disinformation operation is not a conventional propaganda campaign aimed at changing minds; it is an environmental strategy aimed at degrading the conditions under which democratic deliberation can take place. As the 2027 campaign intensifies, the question remains whether French institutions can respond effectively. The detection systems, the legal framework and the investigative responses are all necessary, but they may not be sufficient. The resilience of the system will depend not only on the government but also on media organizations, political parties and voters themselves. If the public begins to treat every claim, every video and every official statement as potentially fabricated, then the disinformation networks will have achieved their goal regardless of whether any single lie is believed. The 2027 election will therefore be a test not just of France’s ability to defend itself against foreign interference, but of its ability to preserve a shared basis of fact in a digital environment designed to destroy it.