The Kremlin's Quiet Hand on the Ballot
Ahead of Germany's February 2025 federal election, the ratings service NewsGuard catalogued 22 false claims, some traced to Russian-linked operations, that appeared designed to boost the far-right Alternative for Germany. Separate analyses found that platform recommendation systems were disproportionately serving pro-AfD content to ordinary users.
On Feb. 23, 2025, Germany held a snap federal election. In the weeks before the vote, disinformation researchers documented an unusually organized effort to shape the online conversation, part of it linked to Russia and much of it favorable to the anti-immigration Alternative for Germany party, known as the AfD.
NewsGuard, which rates the reliability of news sources, said it had identified 22 false claims circulating around the election, including narratives spread by Russian actors targeting mainstream parties that support NATO and Ukraine and appearing designed to boost the AfD. Among them were claims that a government was planning to harass or kill far-right politicians and that ballots had been altered to disadvantage the party.
A network built to look local
Much of the Russian-linked activity, researchers said, ran through fake news outlets. NewsGuard attributed a network of 102 websites designed to masquerade as independent German local news sites, with names such as “Berliner Tageblatt” and “Hamburger Anzeiger,” to John Mark Dougan, a former Florida sheriff's deputy who fled to Russia in 2016. One falsehood seeded on the network claimed that Germany planned to import 1.9 million Kenyan workers, a narrative aligned with the AfD's anti-immigration messaging.
Other fabrications used manipulated video. NewsGuard and other analysts documented a deepfake falsely claiming that Vice Chancellor Robert Habeck had conspired with Ukraine to steal paintings from a Berlin gallery, and another falsely implicating the conservative candidate Friedrich Merz in a decades-old case. Researchers tied several of these to Storm-1516, a Russian influence operation that Microsoft had previously tracked targeting the U.S. presidential election.
Official warnings
German authorities issued public alerts in the campaign's final days. An Interior Ministry spokesperson said that “a targeted campaign is currently being used to try to influence the federal election campaign,” according to Recorded Future News. The spokesperson said a video purporting to show the destruction of ballot papers had been attributed to Storm-1516 and described by U.S. services as a Russian influence operation. The Federal Office for the Protection of the Constitution had warned for months about Russian interference attempts, the outlet reported.
Feeds tilted toward the far right
As the false claims spread, separate research suggested that platform design was compounding the imbalance. The advocacy group Global Witness set up test accounts on TikTok, X and Instagram that followed Germany's four largest parties equally, then measured what the algorithms recommended. On TikTok, 78 percent of the recommended political content from accounts the users did not follow was supportive of the AfD, and on X the figure was 64 percent, according to the analysis published Feb. 19, 2025. The group found no comparable bias on Instagram.
Ellen Judson, a senior campaigner at Global Witness, said the source of the tilt was unclear. “One of our main concerns is that we don't really know why we were suggested the particular content that we were,” she told TechCrunch, adding that her “best inference is that this is a kind of unintended side effect of algorithms which are based on driving engagement.” TikTok disputed the methodology, saying a small number of test accounts did not reflect how its personalized feed works, and X did not respond to the group's findings.
Together, the two strands — coordinated false claims and algorithmic amplification — illustrate a pattern that researchers and public bodies across Europe have flagged repeatedly: that foreign influence operations and engagement-driven recommendation systems can reinforce one another during elections, and that both remain difficult to measure and to regulate.