Dozens of national, regional, and local elections are on the calendar across Europe in 2026, but the messages reaching voters are increasingly synthetic, created by machines, and verified by no one.<\/p>\n
AI-made faces<\/a>, staged crises<\/a>, and nostalgia built for engagement<\/a> now circulate as standard campaign material around the bloc. In some cases, foreign actors<\/a> also interfere in national elections and deploy artificial personas to steer voters.<\/p>\n Brussels has tried to keep up by building the world\u2019s toughest digital rulebook,<\/a> the Digital Services Act, to counter this threat. Yet as elections accelerate, lawmakers are learning that writing law is easier than governing the algorithm\u00a0\u2013\u00a0and that AI-driven politics are spreading\u00a0at a pace hard to keep up with.<\/p>\n Deepfakes, slop, and GenAI\u00a0<\/p>\n Across Europe\u2019s elections, generative AI, \u201cAI slop,\u201d and deepfakes now form three stacked layers of digital political influence.\u00a0<\/p>\n Deepfakes \u2013 manipulated audio or video falsely attributed to real figures\u00a0\u2013\u00a0remain the high-risk, low-volume threat, flaring briefly in places like\u00a0the 2025 <\/a>Irish presidential elections.\u00a0<\/p>\n AI slop<\/a> does the dirty work: the industrial production of cheap, emotionally charged synthetic images and videos dramatizing migration, crime and social decay. This content now dominates recommendation feeds, particularly on TikTok, driven largely by monetised AI-only accounts rather than official campaigns.\u00a0<\/p>\n In this ecosystem, generative AI \u2013 artificially created text, imagery and music from basic prompts \u2013 functions as a multiplier, accelerating polarising narratives faster than Europe\u2019s regulatory machinery can respond. \u00a0<\/p>\n The use of AI exists across the political spectrum. Still, the far right\u2019s political vocabulary<\/a> \u2013 words like identity, threat, cultural siege \u2013 maps almost perfectly onto what algorithms reward: repetitive and provocative content.<\/p>\n AI manipulation across the bloc<\/p>\n Evidence of AI\u2019s impact on elections is already emerging. During France\u2019s national elections in 2024, researchers at the non-profit AI Forensics identified roughly 60\u00a0fully AI-generated political posts<\/a> on official party accounts across the French political spectrum, mainly on Instagram and Facebook. <\/p>\n That filled social media feeds with dystopian skylines, collapsing cities, and hyper-real migration scenes designed to be felt, not verified. <\/p>\n \u201cIt\u2019s less about persuading people what is fake or real\u2026 It\u2019s more about getting an emotional response. Fear, scandal, shock,\u201d Natalia Stanusch, a researcher at AI Forensics, told Euractiv.<\/p>\n In Germany, candidates largely avoided using generative AI in the 2025 federal elections, but TikTok supporter networks embraced synthetic nationalist visuals.<\/a> In Hungary, which is preparing for a high-stakes ballot in 2026, fake profiles<\/a> of attractive young people have been feeding users pro-government messaging.<\/p>\n Moldova, meanwhile, offered a preview of full-spectrum foreign AI interference. Russian-linked networks deployed synthetic TikTok \u201cgrandmothers\u201d<\/a> urging voters to back Kremlin-approved candidates, according to a report published by the Romanian journalism centre Mediacritica, which tracks disinformation.\u00a0<\/p>\n Transparency<\/p>\n A few years ago, Brussels feared a deepfake that could topple a government overnight. That moment never came. What did arrive was much harder to fight: large-scale AI propaganda<\/a>.<\/p>\n On paper, the EU has built a serious defence. The AI Act<\/a> classifies political manipulation as a \u201chigh-risk\u201d activity, requiring transparency and oversight by national authorities and the new European AI Office. The\u00a0Digital Services Act<\/a> requires the largest platforms to reduce election-related risks. Other EU plans and codes add rules<\/a> for political ads and online disinformation.<\/p>\n In reality, major platforms such as TikTok retain broad discretion to determine what qualifies as manipulation, giving them significant control over how the rules are applied.<\/p>\n There is no reliable way to measure how much AI-generated content circulates undetected or goes unreported, and enforcement remains the system\u2019s weakest link.<\/p>\n (mm, cm)<\/p>\n