FOLLOW US

Ukrainian strikes on Russian data centers have fueled claims that Moscow’s propaganda accounts suddenly disappeared from social media. The outages are real. However, the traffic measurements behind those claims do not establish that political bot networks went silent.

Across social media, users have described quieter comment sections and fewer accounts promoting pro-Trump or other Kremlin-friendly narratives. Those observations have raised a broader question: How much of the political anger Americans encounter online comes from people who are pretending to be someone else?

The question deserves scrutiny. Government investigations, platform disclosures and academic research provide substantial evidence of covert efforts to manipulate online discussion. They also reveal important limits on what investigators can conclude about those operations’ reach and effectiveness.

The emerging Yandex story brings those issues together. It offers evidence of major infrastructure damage, alongside a viral interpretation that runs ahead of the available data.

Understanding the difference is essential to identifying manipulation without spreading another unsupported narrative.

What happened to Russia’s data centers

Yandex, the Russian technology company known for its search engine and cloud services, confirmed attacks on two facilities in an Oct. 9 company statement.

The company said its Sasovo data center suffered serious damage overnight Oct. 7-8 and shut down completely. Another attack struck its Kaluga facility on the morning of Oct. 9, disabling part of its infrastructure. Yandex reported no employee injuries at either location.

A third facility, in the Vladimir region, shut down after an attack reported Oct. 11. Reuters reported that Yandex acknowledged damage and service interruptions there as well.

The attacks occurred during an expanding campaign against digital infrastructure. Speaking to reporters Oct. 8, Ukrainian President Volodymyr Zelenskyy described Ukraine’s actions as a response to Russian attacks on Ukrainian data centers.

“We always respond in mirror-like fashion,” Zelenskyy said, without specifically identifying the Yandex strike in those remarks.

Those events establish a wartime infrastructure story. Determining whether they also disrupted covert political operations requires evidence connecting those operations to the affected facilities.

What the Russian bot traffic chart actually measures

An Oct. 11 Tom’s Hardware article reported that automated traffic from Yandex’s network had fallen to 0.4% of its recent peak. The article cited Cloudflare Radar, which tracks internet traffic.

However, “bot” has a broader technical meaning than the one commonly used in political arguments.

The Cloudflare chart measures automated web requests associated with a specific Yandex network, AS13238. Cloudflare’s classification estimates whether requests are automated or human-generated. It does not identify the requests’ political purpose or count social-media accounts.

Yandex’s technical documentation identifies AS13238 as one of the networks used by its web crawlers. These programs retrieve pages for search indexing and other services. Cloudflare separately lists YandexBot as a verified search crawler.

Consequently, a collapse in automated requests from that network cannot be translated into a matching collapse in propaganda accounts. Nor does the reported percentage describe the share of Russian political bots eliminated worldwide.

The distinction does not establish that every affected automated process was harmless. It establishes the limits of the measurement.

A chart showing fewer web requests cannot, by itself, identify the missing users in someone’s comment section. That connection requires account histories, posting data and evidence of who controlled the accounts.

Russian influence operations are already documented

The uncertainty surrounding Yandex does not erase the record of Russian efforts to influence Americans.

In its bipartisan report released in October 2019, the Senate Intelligence Committee found that the Internet Research Agency sought to help Donald Trump and damage Hillary Clinton during the 2016 election.

The committee linked that effort to Kremlin direction. It also found that the operation exploited existing divisions involving race, immigration and gun rights, with African Americans receiving particularly intensive targeting.

Those findings describe an operation that worked through Americans’ existing concerns and conflicts. The operators did not need to invent every grievance to exploit it.

The Senate also documented efforts to draw unwitting Americans into real-world activity, including organizing events and obtaining campaign materials. The manipulation extended beyond anonymous comments on a screen.

Still, those historical findings cannot identify an unfamiliar account posting today. A pro-Trump message is not, on its own, evidence of Russian control. Attribution depends on evidence about the operator and the network, rather than agreement or disagreement with the message.

Fake Americans, manufactured identities

A separate case illustrates how investigators have identified actual social-media bot operations.

In July 2024, the Justice Department announced the seizure of two domains and the search of 968 social-media accounts associated with an AI-enhanced Russian bot farm.

According to the supporting affidavits, the operation created fictitious profiles that often purported to belong to Americans. Operators used those identities to distribute messages supporting Russian government objectives. Prosecutors described involvement by personnel connected to RT and Russia’s Federal Security Service.

One example involved a supposed Minneapolis resident sharing a video of Russian President Vladimir Putin. The department said X suspended identified accounts, including accounts it had already detected before the government action.

The alleged deception was embedded in the identity delivering the message. A reader could encounter Russian state messaging presented as the opinion of an ordinary American.

That is a more concrete basis for discussing online psychological manipulation than a sudden change in the tone of a feed. Investigators identified accounts, infrastructure and alleged operators.

How propaganda borrows credibility

The alleged Doppelganger operation demonstrates another method: making propaganda resemble independent journalism.

In September 2024, the Justice Department announced action against 32 internet domains. An FBI affidavit alleged that Russian companies, working under Kremlin direction, used the sites to influence foreign audiences.

Some domains imitated legitimate news organizations. Others operated under their own media brands. According to prosecutors, fake social profiles, advertisements and influencers helped direct people toward the content.

The alleged objectives included weakening support for Ukraine and influencing voters. The Justice Department said investigators obtained internal planning documents, strategy notes and project proposals.

This approach attempted to borrow credibility twice: from the apparent news outlet publishing the material and from the apparent citizen sharing it. Readers could encounter a manufactured message without seeing its government sponsor.

The case also illustrates why the word “bots” can oversimplify the problem. The alleged operation combined deceptive websites, social profiles and distribution tactics. Counting automated accounts alone would not describe the entire effort.

An influence campaign is not proof of persuasion

Evidence that a campaign exists does not establish how many people believed it or changed their behavior.

In a review of the 2024 global elections, Meta said Russia was the largest source of covert influence networks it had disrupted since 2017. The company reported 39 Russian-origin networks during that period.

However, Meta also said most of the covert networks it removed struggled to attract authentic audiences. Some purchased or manufactured engagement to appear more popular. Those findings describe the company’s own detection and enforcement record, rather than a complete inventory of foreign influence online.

Independent research provides another caution.

A 2023 study published in Nature Communications examined exposure to Russian influence accounts during the 2016 election. Researchers linked survey responses with participants’ Twitter feeds.

Exposure was heavily concentrated: Within the study, 1% of users accounted for 70% of exposures. Researchers found no meaningful relationship between that exposure and changes in political attitudes, polarization or voting behavior.

That finding does not prove every influence operation fails. It concerns a particular campaign, platform and sample. Nevertheless, it cautions against treating the existence of propaganda as proof that it determined an election.

These are separate questions: Who attempted manipulation? Who encountered it? What effect did it have? Each requires evidence.

Algorithms also shape the political environment

Covert accounts are only one part of the online experience. The systems selecting which posts people see can also affect political exposure.

A study published in Nature in February 2026 examined a seven-week experiment conducted on X in 2023. Researchers randomly assigned participating U.S. users to algorithmic or chronological feeds.

Among participants who moved from chronological to algorithmic feeds, researchers found shifts toward more conservative positions on several political measures. These included views involving investigations into Trump and the war in Ukraine. Switching in the opposite direction did not produce comparable effects.

However, the researchers did not find significant changes in self-reported party identification or hostility toward the opposing political party. They also warned that their findings were specific to the platform and period studied.

The study does not explain this week’s reported feed changes. It does demonstrate why visibility, account authenticity and political influence must be examined separately.

An account can be genuine while receiving algorithmic amplification. A deceptive account can attract little attention. Neither possibility can be resolved by counting angry replies.

Where Trump’s diesel deal fits

Some social-media users have connected the Yandex attacks to Trump’s agreement with Putin on Russian diesel supplies. The verified chronology does not establish that explanation.

The initial Sasovo attack occurred overnight Oct. 7-8. Trump announced the diesel agreement Oct. 9, after the first attack and the start of the documented infrastructure disruptions.

Reuters reported that the administration temporarily eased sanctions on Russian diesel as Trump promoted additional supplies for American and global markets. Zelenskyy criticized the agreement, warning that it would benefit Moscow’s war effort.

The deal is relevant to the wider dispute among Washington, Moscow and Kyiv. However, the timing does not support presenting the initial data-center attack as retaliation for its public announcement.

Likewise, no evidence presented in the cited traffic measurements connects the agreement to political accounts allegedly disappearing from social media.

Related coverage: Trump Turns to Russian Diesel as Washington Eases Sanctions

What would establish a connection?

A credible investigation into the reported disappearance of political accounts would begin with identifiable accounts and their posting histories.

Researchers would need to compare activity before and after the strikes, using consistent time periods and explaining how the accounts were selected. They would also need to distinguish stopped activity from reduced visibility in individual users’ feeds.

Next comes attribution. Evidence that an account stopped posting would not, by itself, establish Russian ownership. Evidence of Russian ownership would still leave the question of whether its operation depended on the damaged infrastructure.

Platform disclosures, authenticated operational records or independently verifiable network research could help establish those links. The strongest findings would make their methods clear enough for others to examine.

Until then, reports of quieter feeds remain observations worth investigating. They do not provide a reliable count of propaganda accounts or a confirmed explanation for their absence.

The question behind the “digital psyop”

The phrase “digital psyop” captures a legitimate concern: People want to know whether apparently spontaneous political conversations are being covertly engineered.

The public record supports scrutiny of that concern. Investigators have documented fabricated identities, disguised sponsorship and efforts to exploit existing divisions. Academic research also shows why claims about their effects require careful testing.

However, treating every hostile account as a foreign operative would create another distorted view of public life. Genuine political disagreement persists alongside deception, and distinguishing the two takes more than recognizing an objectionable opinion.

The Yandex outages have renewed an important demand for transparency. Users deserve evidence about who is addressing them, which activity is coordinated and how platforms distribute it.

The damaged servers are real. Whether they silenced the political accounts people describe remains an open question.

The documented history of covert influence is serious enough to demand accountability without claiming that one traffic chart has finally explained the internet.