For weeks, a story raced across the internet with the kind of irresistible clarity that modern media consumers have come to crave: Ukraine’s audacious drone strikes on Russian tech giant Yandex had not only disrupted ordinary internet services, but had secretly shredded the very digital arteries powering pro-Russian bot armies supposedly boosting Germany’s far-right AfD party and Donald Trump’s MAGA movement. It was a narrative of poetic justice—technology built by the enemy turned into a weapon that ultimately crippled their own propaganda machine. The tale spread with viral fury, embraced by pundits and even echoed by Ukrainian President Volodymyr Zelenskyy. Yet when researchers peeked under the hood, the story began to resemble something far less dramatic. Fresh investigations published in mid-October are now systematically dismantling the claim, revealing that the purported collapse in political bot activity was largely a mirage—a case of compelling anecdote and sloppy data analysis getting well ahead of verifiable fact.
The most detailed rebuttal came from Danish analyst Valdemar Osted of Common Consultancy, who methodically tested the bot-collapse theory against hard Facebook numbers. Osted’s team examined the social media accounts of 76 AfD politicians and 136 MAGA figures, comparing comments on new posts against each account’s typical baseline performance while carefully adjusting for variables like posting time, day of the week, and the age of individual posts. The results were remarkably unremarkable. After the Yandex strikes, AfD comment levels wobbled between just eight percent below expectations and five percent above them—hardly the catastrophic silence that viral charts promised. MAGA accounts, meanwhile, showed activity that actually exceeded expectations by anywhere from four to ten percent. Far from disappearing into the digital ether, comments on new MAGA posts surged to between 480,000 and 560,000 per day in the aftermath compared to roughly 370,000 in late September. Osted described these variations as completely ordinary fluctuations, the kind of noise that any active political community produces from one week to the next, and the kind of evidence that rarely makes for a splashy headline but holds up to closer scrutiny.
European researchers monitoring suspected influence operations arrived at similarly deflating conclusions. Richard Schwenn, an analyst at the German think tank Polisphere, told Capital magazine that his organization’s surveillance of nearly 10,000 accounts across TikTok, Telegram, and X—a broad net that included known disinformation vectors, AfD-affiliated personalities, and pro-Russian sources—detected no statistically meaningful decline in activity. Their data, collected in partnership with the Hasso Plattner Institute, showed no commensurate drop in either existing bot chatter or the creation of new automated accounts. Josef Holnburger, managing director of the German disinformation research organization CeMAS, likewise reported no related downturn in the Russian disinformation campaigns his team monitors, despite the destruction of the data centres. Poland’s state research institute NASK similarly scrutinized X accounts that it previously flagged for spreading disinformation between September 30 and October 11, and found no noteworthy overall decline after the Ukrainian strikes. While some individual accounts did go quiet, investigators were unable to establish any causal link between their silence and the Yandex attacks. For researchers who have spent years studying the sluggish and often maddeningly resilient machinery of online propaganda, none of this was surprising.
The actual strikes themselves were real enough. Kyiv targeted three Yandex data centres in Sasovo, Kaluga, and the Vladimir region between October 8 and 11, facilities that underpin Russian cloud services, business operations, and artificial intelligence systems. The resulting outages disrupted Yandex services and many dependent websites, prompting Kremlin spokesman Dmitry Peskov to acknowledge that the government was actively discussing the problems and that Yandex was engaged in urgent recovery efforts. But satellite damage on the ground does not necessarily translate into the apocalyptic collapse of political interference online. The confusion arose, in large part, from a crude misreading of Cloudflare data, which recorded automated requests from Yandex’s network plummeting to just 0.4 percent of the previous peak. The vital detail lost in translation, however, was that this metric measured automated web traffic—mostly search-engine crawlers indexing sites—not propaganda accounts posting on social media platforms. When analysts looked at broader Cloudflare figures for automated activity originating from Russia as a whole, there was no comparable collapse. A plummeting metric from one company’s server network simply told us that Yandex’s servers had been damaged, which we already knew; it told us nothing about whether German voters were being shielded from bot-driven manipulation.
Despite these inconvenient realities, the claim multiplied like a well-engineered piece of misinformation itself. A pivotal moment came when Czech MEP Tomáš Zdechovský shared a chart suggesting a 56 percent decline in online activity surrounding the AfD, a visual that instantly went viral. The graphic was a masterpiece of persuasive presentation but lacked any identifiable source, methodology, or underlying dataset. German fact-checkers soon uncovered glaring chronological inconsistencies, including figures that extended to October 15—several days after the chart had already been circulating online, a dead giveaway of hasty fabrication. Polish military commentator Jarosław Wolski helped amplify the claim that AfD-related bot activity had fallen by 56 percent while also hinting at an abrupt decline in online MAGA support. His posts reached millions before he deleted them on Tuesday and offered a public apology for spreading false information. Poland’s intelligence services coordinator, Tomasz Siemoniak, separately claimed that the domestic security agency ABW had observed a major decline in Russian bot activity, though when pressed, he clarified that he had relied on ABW information rather than the viral graphic—an awkward distinction that nobody fully explained.
The episode reached its most uncomfortable moment when Ukrainian President Zelenskyy suggested on Monday that social media had grown “less hostile” following the strikes, attributing this to fewer bots. It was an appealing claim for a wartime leader, yet Ukraine’s own Center for Countering Disinformation was forced to walk it back the following day, admitting it had no evidence of a major overall reduction in Russian bot-network activity. The center reported only narrower, temporary disruption to specific networks like Pravda, InfoDefense, and Nod of Time, with some clusters showing posting halved for roughly twelve hours before recovering. That, in the end, is what actually happened: a brief interruption, not a strategic triumph. The broader lesson here is both old and uncomfortable. In an age where speed of sharing matters more than depth of verification, a compelling chart and a heroic narrative can easily travel around the world before the careful researchers arrive with their spreadsheets and methodological caveats. The desire to believe that the enemy’s propaganda machine has been decisively broken is powerful, particularly when the news cycle otherwise offers little good news. But the deflating, humanizing truth is that propaganda networks are resilient, distributed, and woven into the broader fabric of the internet—they can absorb damage, adapt, and recover. The battle against disinformation is not won by a single strike, no matter how precise or satisfying it might feel.