One network, three elections
In October 2025, Dutch researchers mapped more than 550 fake accounts on X. The accounts boosted right-wing, anti-immigration, pro-Russian and anti-Ukrainian posts before the Dutch parliamentary election.
In April 2026 the same network turned up around Hungary’s vote. In September 2026 it turned up around Sweden’s general election. Some accounts changed their names and photos between campaigns. Many kept the same habits.
On 16 November, EUobserver’s article on the latest findings described them as “at least 8,505 Russian fake accounts.” That label does not match what the researchers wrote. The difference is the main subject of this article.
What the researchers found
The Swedish findings come from Alliance4Europe and the Dutch firm Trollrensics, working with Swedish broadcaster SVT. Their report says they identified 8,505 accounts consistent with the network’s markers. They say this is not an exhaustive count.
Most of the Swedish-focused accounts are listed on X as based in Benin and Nigeria. Others are listed in Côte d’Ivoire, Ghana, Hong Kong, Vietnam, Indonesia and Malaysia.
The accounts post little. They follow a lot. In a 24-account sample, 14 followed almost exclusively Swedish users. One account, with a Dutch bio, changed its name and photos on 17 July 2026. It now follows 750 Swedish and 766 Dutch accounts.
Euronews reports that some Swedish profiles with thousands of followers drew as much as 30% of those followers from suspicious accounts. The followed accounts leaned right, especially figures linked to the Moderate Party. A few were left-leaning or neutral. The report says it does not allege that any followed person knew about the operation.
The tactics
1. Borrowed local identity
The accounts used localised names and handles and stolen profile photos. The Dutch phase also used photos of older women in hospital settings. The goal is to look like an ordinary Swedish or Dutch user. A stranger with an Indonesian profile would not carry weight in a Swedish political thread.
2. Follow-based amplification
Mass following does two jobs. It builds a follower base that looks real. It also puts these accounts in the middle of Swedish political conversation. Alliance4Europe says X caps new accounts at 7,500 follows, yet many of these accounts exceed that several times over.
The followed accounts gain from it too. Researchers quoted by Euronews say mass following can inflate popularity, and that coordinated likes and reposts can then push posts further.
3. Reuse across campaigns
A network built for the Netherlands was reused in Hungary, then Sweden, by renaming accounts and swapping photos. The Swedish phase began with Dutch and Hungarian assets following Swedish accounts. The authors used that overlap to link the three.
Reuse is cheap. It is also easier to detect, because the same behaviour repeats. Alliance4Europe says some accounts were flagged as early as 2024. Trollrensics uses 26 indicators, including spikes in account creation on specific dates, stolen photos and identical posts published at the same moment.
4. Plausible deniability
The Dutch-phase content leaned pro-Russian and anti-Ukrainian. That invites the assumption that Russia ran it. The researchers did not say so.
Robert van der Noordaa, co-founder of Trollrensics, told Euronews he believes the accounts are part of a commercial “guns for hire” operation. Euronews presents that as his belief, not a finding. Darren Linvill of Clemson University’s Media Forensics Hub examined a sample. He said the behaviour appeared consistent with paid amplification services, and compared the methods to those seen in Russian operations. Euronews did not attribute the accounts to Russia.
A paid amplification service sells reach to whoever pays. A buyer could be a party, a foreign state or a company. Nobody has shown who paid here.
What the Influence Tactics Score would show
This is my reading of other researchers’ published data. It is not a scored analysis of the content itself.
- Synthetic identity: High. Stolen photos, localised names, repurposed accounts.
- Coordinated amplification: High. Mass following, identical posts, simultaneous activity.
- Tribal division: Moderate to high in the Dutch phase, given the anti-immigration framing. The Swedish accounts mostly followed rather than posted, so there is little content to score.
- Missing information: High. Readers see a follower count and a profile photo. They do not usually open each follower’s join date or listed country, or see the account’s earlier campaigns under other names.
How EUobserver relabelled the network
EUobserver reports “at least 8,505 Russian fake accounts.” The Alliance4Europe report calls them Benin-based accounts consistent with the network’s markers. It does not name a sponsor.
Alliance4Europe does document a separate Russian operation in the same period. Its Matryoshka report describes at least six fake news videos published on 18 August. Each impersonated a different outlet. X removed that content about six days after publication.
That operation uses impersonated media brands and is attributed to Matryoshka, a Russian operation Alliance4Europe has tracked before. The follower network uses stolen faces and mass following. Merging them in one news story makes the story simpler. It also makes it less accurate, and it gives critics of the research a factual error to point to.
What this is not
This does not show that the accounts changed any vote. The researchers did not measure that, and their samples are not statistically representative. It also does not show that any party or government commissioned the network. The report names no buyer.
Limits of this analysis
I read the Alliance4Europe network report and its 16 November brief directly. I read the Euronews and EUobserver stories as press coverage. I did not examine the account data.
The two Alliance4Europe pages give slightly different timing for the 8,505 count. The network report also spells some handles inconsistently. I could not confirm how Alliance4Europe or Trollrensics are funded. Treat the findings as strong leads from researchers working with newsrooms, not as court-tested facts.
The Euronews story reports no response from X. Alliance4Europe says fewer than 1% of the accounts it logged earlier have been restricted, removed or banned. I found no statement from X on that figure. X is also the only party that can confirm what it removed, and it has a stake in the answer.
What to watch for
You can check a few things yourself.
- Follower quality. Open a political account’s followers. If many have generic names, no posts and a country that does not match the topic, the audience may be bought.
- Profile history. Look for name or photo changes. An account that discussed Dutch politics last year and Swedish politics now has a past that does not match its bio.
- Attribution claims. When a headline names a state, find the sentence in the underlying report that does. If it is missing, the claim is the journalist’s inference.
- Next target. Alliance4Europe notes two accounts already following Serbian accounts ahead of Serbia’s next election. If the pattern holds, the same accounts will appear there.
This article is part of Decipon’s Manipulation Breakdowns series, examining specific influence operations through the Influence Tactics Protocol.
Sources:
- From the Netherlands to Sweden: Guns For Hire CIB Network’s Third European Election Cycle — Alliance4Europe
- State of Manipulation: A threat to Europe’s security — Alliance4Europe, 16 November 2026
- Illegal Russian influence operation continues to go unimpeded on X: Matryoshka targeting the Swedish Elections — Alliance4Europe
- Thousands of suspicious accounts followed Swedish political voices, researchers say — Euronews, 16 September 2026
- More EU sanctions on Russian propagandists and hackers, as thousands of fake accounts target voters — EUobserver, 16 November 2026