On January 27, 2025, less than a month before the German federal elections to the Bundestag, a X account announced that explosives had been placed next to several polling stations. Six minutes later, the message had already been reshared 202 times, all within the same minute. The Institute for Strategic Dialogue traced the thread and found that 48 accounts were fabricating the stories while another network, this one with more than 6,000 profiles, was moving them. They used logos of Deutsche Welle, the BBC, Sky News, and, atop real videos of the British police, added AI-generated voices that did not utter a single word in German. No one intended to sway a vote that night; it was about inserting noise, manufacturing an emergency out of thin air, and forcing institutions, media, and citizens to chase after an event that did not exist. For a few hours, the attacker decided with which lenses to read the electoral campaign, and there was no need to touch a ballot.
There begins a problem that Spanish voters will face in 2027. A foreign power seeking to interfere in those elections would have little reason to invent a society different from the one we already have: we are already a country torn by debates strong enough about immigration, housing, corruption, inequality, territorial identity, or how institutions function. It is enough to poke the sore spot that was already bleeding. In these cases, the most effective influence operations do not invent a fracture because it is unnecessary. They exploit the fractures that exist, study them, learn the language of each community, and let national politics do the rest.
Because the fundamental mistake—one we repeat election after election—is to talk about this as if there were a single enemy with a single intent, when reality shows that multiple actors with agendas that hardly align compete, clash, and sometimes even profit from one another’s actions, as long as the outcome is what they desire. After Valencia’s October 2024 DANA, a China-linked operation named Spamouflage disguised itself as the Spanish NGO Safeguard Defenders to urge, on Facebook, TikTok, and X, the overthrow of the government. This operation managed to get substantial numbers of Spanish users to share those accounts without knowing what state machinery lay behind them, and for days the call for resignation circulated mixed with legitimate anger, until part of the reasoning was forgotten. There you see how the strategy is not to lie about a fact, but to hijack the recipient’s emotional state until anger replaces reasoning.
“The strategy is not to lie about a fact, but to hijack the recipient’s emotional state until anger replaces reasoning”
At the end of July this year, the Ceuta crisis illustrated why it is wise to separate proven facts from suspicion. The National Center for Immigration and Borders documented that the mass crossing of tens of thousands of people responded to three manipulated factors. First, a simplistic interpretation of a Supreme Court ruling summarized on networks suggesting that those who crossed by swimming could not be returned. Second, the idea that the Moroccan police apparatus, deployed at the Throne Celebration on July 30, would leave the border unprotected. Third, the rumor of an extraordinary regularization in Spain that did not apply to those who crossed on those days but circulated like wildfire. Hence its large impact: it does not invent a fact out of nothing; it distorts a real one until it ceases to mean what it meant. What followed (pro-Israeli accounts calling Ceuta the “Spanish colony,” Pravda reusing a 2025 photo as if recent to accuse Spanish ministers) were also informational attacks on other past events. In Ceuta, cognitive attacks were causes and consequences.
To sustain manipulation, first impressions matter, but they must be supported by more complex networks across the information chain as people decide what to learn. In August 2026, The Guardian documented that the supposed Hanover Institute had published 124 reports in nine days. Its distributor testified to American authorities that it worked for the Israeli government. Although the page carried a notice of foreign ties, the texts were presented as neutral research and were tailored to draw quotes from AI assistants and poison them. With this method, the aim was to place material where a reader looks for documentation to dispel doubts.
In this sense, interference takes a form less visible than traditional propaganda. A foreign actor may not have the clout to impose a view, yet they possess enough power to decide which topic sits at the center of conversation. If a single issue dominates a whole week of campaigning, the conditions under which parties compete are radically different. Why? Because it compels replies, displaces other debates, and favors whoever finds a roomier path in that terrain. The cognitive battle begins long before deciding what someone thinks; it begins by deciding what they will think.
“The cognitive battle begins long before deciding what someone thinks; it begins by deciding what they will think”
For that reason, seemingly clumsy operations can turn out to be effective. The European External Action Service documented 540 information-manipulation incidents in 2025, with 27% already supported by AI-generated content, nearly triple the previous year. Much of the material remained mediocre, with little genuine engagement, but that hardly matters. AI lowers the cost of testing a hundred stories at once to see which one a society is willing to adopt. Elsa Hedling, in her study for the Hybrid CoE, explains how uncertainty intensifies the search for shared explanations within each community and how a falsehood finds fertile ground when it seems to explain why some expect aid while others supposedly enjoy a privilege. The core issue remains the same, but the organizing question shifts: the lack begins to be explained by the existence of an invented beneficiary.
The core question remains almost always the same, and that is why it is important not to lose sight of it: exploiting preexisting fractures—such as territory, migration, housing, or the generational gap—because misinformation works best when it stirs an existing division rather than inventing a new one. The how evolves: it is no longer a stray rumor, but a three-tier production chain, with institutional content at the top, intermediaries translating and giving it a local voice in the middle, and channels delivering it to social groups at the bottom, so that the message never arrives labeled as foreign propaganda, but dressed in its own indignation. The why is not limited to winning a particular election; it concerns ensuring that a member state of the EU and NATO is as divided, distrustful, and unstable as to hinder unified action when necessary and as confused as to allow its own institutions to introduce more noise than they remove.
Hannah Arendt wrote that organized lying does not aim to make people believe something false, but to prevent them from distinguishing truth from falsehood, and that diagnosis, more than sixty years old, describes what we now see on Spanish networks. Because, in the end, neither Russia nor Morocco nor China need us to vote for what they want; for them it suffices that we stop debating the facts and begin distrusting those who should clarify them.
“Disinformation works best when it erodes an existing division rather than inventing a new one”
With these ingredients, interference in the 2027 general elections should not surprise us. Three lines loom clearly ahead. First, the actors will multiply—Russia, Morocco, China, Israel, and national players seeking their own gains. Second, the technique keeps improving; these operations no longer seek virality but aim to exhaust fact-checkers by flooding them with synthetic content until they can no longer keep up, and that same recipe fits the Spanish electoral cycle. Third, attribution will become increasingly deliberately ambiguous. For instance, in Ceuta no one could fully prove what was spontaneous and what was induced, and that same fog will become a domestic political weapon in 2027, because accusing a rival of “foreign interference” will be more electorally profitable than proving it.
One can expect electoral hoaxes synchronized with real crises (another blackout, another flood, another wave of migration) where the event itself supplies fear and disinformation supplies the culprit, false bomb threats against polling places generated with synthetic voice as in Germany, and above all, a domestic ecosystem that learns from all these actors without external directions. In a single vote there is more than a name: there is memory, trust, and the country’s version with which one goes to bed that night. What those 2027 elections decide will depend, more than the ballot boxes, on how many of those wounds were truly ours and how many were opened by those who care more about the voter’s interests than the vote.
