Bots, Russian data centres and the illusion of public opinion
Open a political post and scroll through the comments. Strangers repeat the same claim. Hundreds of likes make it look convincing. Before you check a fact, the crowd has suggested what everyone thinks. But how many people are actually there?
That question resurfaced after Ukrainian drone strikes hit data centres belonging to Russian technology company Yandex between 8 and 11 October 2026. Facilities in Sasovo, Kaluga and Vladimir were affected. These buildings house computers that keep websites, apps and other services running. [1]
Soon, stories spread about quieter social media. Posts on Facebook described fewer hostile comments. On X, Garry Kasparov reported fewer accounts he considered Russian trolls or bots. Others noticed less apparently automated support for right-wing causes. These were personal observations, not verified results across either platform. [2] [3]
The widely shared graph showed a drop in activity by automated programs on Yandex’s network. It did not count propaganda accounts on Facebook or X. Those programs also include tools that help search engines find websites. The attacks happened; whether they caused a widespread fall in political bots remains unproven. [4] [5]
A bot is a computer program that performs actions through an account: posting, liking or sharing. Some are useful. Others impersonate people. A troll may be a real person paid to provoke arguments. Campaigns can combine both. [6] [7]
Think of a small team wearing thousands of masks. Operators choose the message. Software manages the accounts, while AI can create pictures, biographies and variations of an argument. Accounts support one another with replies and likes, making organised propaganda look like ordinary conversation. [7]
Real people then join in and spread it further. The apparent crowd helps steer what we notice and consider normal. Meta has found influence campaigns across Facebook, Instagram, X, YouTube, TikTok and other services. Many struggled to attract real audiences, however: trying to influence people does not guarantee success. [8]
Who sets this up? Scammers, businesses selling fake followers and political operators all use bots. In 2024, US authorities described a Russian operation involving 968 accounts on X. Court filings said an executive at Russian state broadcaster RT organised its development; a Russian security-service officer later led the organisation using it, with Kremlin approval and funding. [6] [9]
Dutch intelligence said that campaign ran from a computer in the Netherlands. The organisers, computers and places claimed in the profiles could therefore be in different countries. Russian computers going offline would not necessarily stop Russian campaigns. [10]
Why the connection with the right? Russia did not invent right-wing politics, and conservative views are no sign of a bot. But a US Senate investigation supported by both major parties found that Russia’s Internet Research Agency backed Donald Trump and worked against Hillary Clinton in 2016. [11]
Other documented operations sought to weaken support for Ukraine. My reading is that movements opposing Ukraine aid or cooperation between Western allies offer useful audiences for those goals. Russian operators have also posed as both liberals and conservatives: backing a candidate and encouraging division can happen together. [12] [13]
Picture an invisible hand holding an invisible stick, prodding arguments from a distance. Operators exploit existing fears about immigration, race or identity. Real people supply the anger and carry it into their relationships. Making a society fight itself can serve an adversary’s interests while keeping the instigator hidden. [11]
The danger is becoming harsher towards neighbours and treating disagreement as betrayal. That does not mean every argument has a foreign organiser. Nor is Russia alone: Meta has reported influence networks originating in Iran and China too. [8]
Why do America and Europe want more data centres? They support jobs, research, public services and AI. The US wants technological leadership and stronger economic and national security. The EU wants more computing capacity and less dependence on overseas providers. Building these facilities does not mean building bot farms. [14] [15]
Yet our own centres need scrutiny. They can host bot farms or act as middlemen that hide where activity originates. In March 2026, Dutch police warned that companies buying and reselling computing space could hide criminal services inside legitimate businesses. The Dutch computer used in the Russian campaign shows the political risk too. [10] [16]
The companies renting out computing services should check who they are renting to, provide a way to report abuse and investigate credible complaints. A voluntary Dutch industry code already includes customer checks and action against misuse. Governments can help providers share warnings; police investigate crimes, while intelligence and defence services investigate foreign threats within their legal powers. [10] [17] [18]
Privacy must be part of this from the start. Investigations should follow evidence of abuse, not political opinions or everyone’s private conversations. Providers should collect only necessary information for a lawful purpose, limit who can access it and delete it when no longer needed. European data-protection rules require these principles. Customers should be able to challenge mistakes. [19]
Intelligence agencies operate under separate laws and need independent scrutiny. Dutch watchdogs check permission for certain intrusive powers and oversee their use. In September 2026, they warned that proposed wider powers lacked adequate safeguards. Protecting our computers must not become a blank cheque for surveillance. [20] [21]
Fortunately, hidden coordination leaves visible dots: repeated wording, matching posting times, shared computers and money trails. Good investigators connect them using evidence. Reading between the lines can suggest where to look; it cannot replace checking the facts. [7] [9] [12]
Five clues can help spot a likely bot. Look for several together; disagreement alone proves nothing. [6] [7]
- Check the rhythm. Constant posting at fixed intervals can suggest a computer program.
- Look for repetition. Identical comments across different accounts suggest organised activity.
- Read older posts. Sudden changes in identity or subject deserve a closer look.
- Check the picture. Search the profile photo online: it may belong to somebody else.
- Watch the replies. Repeated off-topic responses or accounts constantly boosting one another are clues.