Ukrainian drone strikes on Yandex data centers in early October prompted a rush to connect the attacks to a decline in Russian propaganda bot activity online. President Zelenskyy himself floated the connection at the Warsaw Security Forum, though his office later walked it back, reframing his comments as observations about social media "atmosphere" rather than measurable bot metrics. The real damage came from a graph — sourceless, undated, and riddled with errors — that spread as supposed proof. Czech MEP Tomáš Zdechovský shared it on X, where it racked up over four million views. He suggested it might show declining "online activity around AfD," though the graph itself contains no reference to AfD, to any specific platform, or to any named data source. The graph's problems are not subtle. Its Y-axis spacing is inconsistent — the distance between 2,500 and 3,000 is visibly shorter than other 500-unit intervals. The Y-axis doesn't start at zero, visually exaggerating the decline. The X-axis represents equal time periods with unequal spacing. Most damning: the graph, which first appeared online October 11, includes data points through October 14 — three days into the future. You cannot chart data that doesn't exist. Josef Holnburger, managing director of CeMAS (Center for Monitoring, Analysis and Strategy), which systematically tracks Russian disinformation campaigns, is unequivocal: there is "no decline in Russian disinformation activities that could be explained by the attack on Yandex data centers." The infrastructure used by disinformation-related social bots is likely distributed across many providers, not concentrated on Yandex servers. Cloudflare data, also cited by some as evidence, reflects a different phenomenon entirely. Cloudflare blocks non-human traffic to the websites it services — search engine crawlers, spam bots, content scrapers. That is "something entirely different from a decline in inauthentic behavior by social bots on social networks," as Holnburger told Germany's #Faktenfuchs. When examining overall activity from Russian IP addresses on Cloudflare-serviced servers, there is no major decline in bot activity. The episode is a textbook case of confirmation bias laundered through data visualization. A fabricated or AI-generated chart, absent any source, methodology, or even temporally possible data, was amplified by an elected European lawmaker to four million viewers because it confirmed a narrative people wanted to believe. The actual experts monitoring Russian disinformation say the premise is wrong. Whether AI generated the graph is unproven but plausible — the "hallucinated" future data and scaling errors are consistent with AI-produced charts. The creator remains unknown. The larger pattern is clear: false or fabricated graphs lend credibility to unsupported claims, and the correction never travels as far as the original.
