We ALL Fell for the Same Russian Bot Chart

By Rachel Hurley October 12th, 2026

Ukraine really hit Yandex. The bots that went quiet were never the trolls.

Welp.

I posted on FB that Ukraine silenced Russia’s bot farms, and the chart I used to prove it was measuring something else entirely.

A commenter told me I was spreading misinformation. I was annoyed for about four minutes, and then I went and checked, and they were right.

The real story is wilder than the one I posted.

Ukraine hit three Yandex data centers in four days, and Yandex is basically Russia’s Google. Sasovo went down October 8, part of Kaluga went down October 9, and Vladimir went down October 11. Zelensky called it payback for Russia bombing Ukraine’s data centers all September.

A bunch of Russian stuff that ran on those servers stopped loading. That part was real.

The part I got wrong was the chart.

It came from a free site called Cloudflare Radar. It showed robot traffic from Yandex’s network falling to 0.4% of its peak. I shared it, and so did everybody else.

Frank Luntz, the pollster you’ve seen on TV, posted that Russian bot traffic had plummeted. Historian Timothy Snyder called it a quick take, but he said it sure looked like pro-Trump and pro-AfD bots run on Russian servers. Tomáš Zdechovský, a Czech member of the EU Parliament, said chatter about Germany’s far-right AfD party fell off a cliff. On Monday, Zelensky said everybody noticed social media got easier to use.

So I was in pretty fancy company.

Cloudflare counts every robot that asks a website for a page as bot traffic. I should’ve known that, cause I use Cloudflare and I’ve spent hours inside it learning how it works.

When you knock out a search company’s data centers, the first thing that dies is its crawler. That’s the little program that visits every website on earth so Yandex search knows what’s on them. If you have a website, you know you can let Google crawl it or turn that off.

What went to zero was a bunch of robots doing chores.

Diagram: Yandex's search robot asks websites for pages, and Cloudflare counts that as bot traffic. A troll account posting on X or Facebook never shows up in that chart.

The group antibot4navalny spends its days hunting Kremlin sock puppets. They told Radio Free Europe the chart “has little to do with malicious propaganda bots on social media.” Russia’s overall bot numbers looked like any other week. By Sunday night, Yandex’s traffic was already climbing back to about a fifth of normal.

So nobody silenced the trolls. The crawler bots just got sidelined.

Disinformation analyst Joohn Choe traced it back, and I followed his trail. Walk through it with me.

On Thursday, October 8, security researcher Jeremiah Fowler posted that Russian bot traffic had dropped to 49%.

On Sunday, October 11, Mario Nawfal, who runs one of the biggest news accounts on X, shared it.

Less than an hour later, radio reporter Jamie Dupree passed it on.

That same day, Tom’s Hardware ran a headline calling it Russian disinformation bot traffic.

By Monday, it was in Le Parisien, one of France’s big papers.

Tom’s Hardware has since changed its headline, but the old wording is still sitting in the web address. Le Parisien’s headline said pro-Trump and pro-AfD bot activity had collapsed. The article underneath used the French verb form for stuff you heard but can’t confirm. The headline sounded sure, even though the article wasn’t.

Le Parisien also said AfD support online fell 56%, with no source. That number came from a chart Zdechovský posted. It’s labeled “daily activity” and has no source. Its scale starts at 2,500 instead of zero, which makes the drop look way steeper.

The Austrian fact-checking site Mimikama spotted something even weirder. The chart’s second low point seems to land on October 13 or 14, which is two days after it was posted. Mimikama called the claim misleading.

A German site called Rohtext tried counting posts that named each German party that weekend. Chatter about every party dropped, and the AfD actually dropped less than the Social Democrats and the Greens. Those counts came from Grok, and nobody could double-check them, so treat them as rough. Still, it looks a lot more like a quiet weekend than an AfD blackout.

Then I pulled a whole month of the same Cloudflare data myself. In the four weeks before the first strike, those robots slowed to about a third of normal four times. Each time, the people using Yandex’s network carried on like nothing happened. The drop Fowler posted about was smaller than all four of those.

Line chart of a month of Yandex's robot traffic. It drops to about a third four times before any strike, falls by half on October 8, and drops to almost zero on October 11.

The rest of what I found was so obvious it’s embarrassing.

Nobody has shown that Russia’s bot farms post from Yandex servers at all. An account posting from a Moscow server would stick out, so the farms rent their disguises from us. They rent other people’s home internet, hack smart gadgets, use VPNs and pay for Western cloud servers. It’s all so an account run out of Moscow looks like a guy in Ohio.

And yeah, there’s receipts.

In July 2024, the DOJ took down a bot farm called Meliorator. An editor at RT, Russia’s state TV network, built it, and Russian spies were in the loop. It ran nearly a thousand fake X accounts pretending to be Americans, including a bitcoin investor in Minneapolis who didn’t exist. The software gave each fake person an internet address that matched where they claimed to live.

The email domains behind those accounts were bought from Namecheap, a company in Arizona. If you wanted to burn down the servers Russia’s trolls actually depend on, you’d have to start a lot closer to home.

I’ll defend myself a little, cause this story was easy to believe for a reason. It’s happened before.

When Israel hit Iran in June 2025, Iran’s internet collapsed, and hundreds of fake “Scottish” accounts went silent at once. Nurses and Glaswegian socialists who’d been posting all day suddenly had nothing to say. When Iran pulled the plug again this year during protests, the same accounts went silent again. Emily Thornberry, a British lawmaker, brought it up in a Parliament committee.

That worked cause the whole country went dark, including the guys at the keyboards. Russia’s internet stayed on, and one company’s search engine got a concussion.

A lot of you pushed back and said you saw the quiet with your own eyes. I believe you saw something. But it was a weekend, and a viral post had just told everybody the bots were gone. Nobody counted the trolls the week before, either.

One commenter said their feeds got quieter, and another said theirs got 15% more bots. Feeds are built for each of us, so two people can see opposite things on the same day.

The best pushback came from a reader who runs web servers. They say political junk has been hitting those servers for years, mostly from US data centers. They say it fell to zero during the second strike.

Those are real logs, so I checked. Cloudflare shows nothing new going dark on Yandex’s networks that day, and nothing went dark until October 11. Spam from US servers would only stop if whoever runs it was sitting on Yandex, and nobody has shown that.

To be fair, Yandex’s whole network went mostly dark early Sunday, and it’s still only partly back. If a troll farm rented space there, this is when it would go quiet. Nobody has shown one did, so I went looking for anyone who’d actually counted.

French researcher David Colon compared two fake stories from a Russian operation called Storm-1516, one posted before the strikes and one after. He says the one after got a few dozen views, when pages like it used to pull millions. That’s interesting, but it’s two posts he picked, and views aren’t accounts.

A Ukrainian analyst who goes by Tatarigami posted that Texas separatist accounts with matching “Veteran, Christian, Father, MAGA” bios had gone quiet. He posted it at 4:33 on a Sunday morning, Texas time. He didn’t post any numbers.

The one group I found that actually counts is CAT-UA, a Ukrainian nonprofit that tracks about 10,000 posts a day. They saw no real drop. That’s a Ukrainian group with every reason to want this story to be true, and they still said calling it would be “too bold.”

Euronews, Radio Free Europe and fact-checkers in Germany and Austria all looked into it too. None of them found evidence that social media bots dropped. Even the Ukrainian tech site Mezha ran a headline saying the strikes didn’t reduce Russian bots.

What caught the Iran network was names and times. Researchers showed which accounts stopped, at what hour, and when they came back. So I ran that same check myself.

I took 26 accounts that antibot4navalny flagged as Kremlin bots on October 10. Six of them never posted that week, so I tracked the other 20 day by day from October 5 to October 12. I checked 15 US and German politicians next to them to compare.

On a normal day, about 11 of the 20 bots posted.

On Sunday, October 11, only 3 did.

The politicians dropped too, from about 12 to 7, cause it was Sunday.

Line chart comparing 20 flagged Kremlin bots with 15 politicians. Both dip on the weekend, the bots dip harder on Sunday, then start coming back on Monday.

So the bots did dip harder than the politicians. But one of them posted right in the middle of Yandex’s blackout, and by Monday six were back while Yandex was still mostly down. A few had gone quiet before the first drone even hit. That’s not what the Iran network looked like.

The bots surprised me in one more way. They’ve got names like Jocelyn Bade and Willow Aldridge, and every post I opened was in Russian. They were snapping at Russian news outlets like Radio Svoboda and the Moscow Times. So these are Kremlin bots working the Russian internet, not the MAGA accounts everybody was talking about.

The people who shared that chart spend all year warning us about disinformation. The second a chart told us a story we liked, we hit share without asking what it measured. I did it too.

Caroline Orr Bueno, a scientist who studies this stuff for a living, basically begged everyone to knock it off. If you want to defend democracy, you have to reject the lies that flatter you too.

A misread crawler chart got more reach in one weekend than most Russian sock puppets get in a year. Real, verified humans with good intentions pushed it. The Kremlin didn’t have to lift a finger. The algorithm rewarded the version that felt good, and we did the rest.

Choe already called the sequel. When Yandex comes back online, somebody’s gonna post that the bots are back, and that’ll be crawlers too.

If I find anything that changes what I’ve found here, I’ll come back and report it.

So thank you to the commenter who called me out.

My bad.

Search every title

We ALL Fell for the Same Russian Bot Chart

Ukraine really hit Yandex. The bots that went quiet were never the trolls.

Sources