The far right in Europe has flooded social networks with AI-generated fake images. From manipulation messages targeting immigration to the idea that we will be forced to eat insects, no theme that can induce fear is avoided by extremist movements.
This year’s European parliamentary elections have served as a launching pad for an AI-generated campaign carried out by the European far right, a phenomenon that has continued to proliferate since then, experts say.
This month, the issue reached Meta's independent oversight board, the company led by Mark Zuckerberg, which opened an investigation into anti-immigration content on Facebook. The oversight commission's investigation will examine a post from a German account featuring an AI-generated image accompanied by anti-immigrant rhetoric.
The post is part of a wave of right-wing content created by AI on social networks, reports The Guardian.
The giant pig chasing Arabs
The German far-right party Alternative für Deutschland (AfD) and its anti-immigrant supporters have extensively used AI image generators. The image analyzed by Meta's content moderation group features a woman with blonde hair and blue eyes raising her hand in a stop gesture, with a stop sign and the German flag in the background.
The text above the image states that people should not come to Germany anymore, as the country does not need more "gang rape specialists" due to the immigration policy of the Green Party.
The Brandenburg branch of AfD produced campaign ads in September made with artificial intelligence, contrasting an idealized Germany with people with blonde hair and blue eyes, and scenes of veiled women walking the streets and a person waving an LGBTQ+ flag. Reality Defender, an American company specialized in detecting deepfake images, stated that the image could be produced in minutes.
Other pro-AfD groups on Facebook use AI-generated images containing chauvinistic slogans. One of them shows a giant pig – an animal whose meat is forbidden for consumption in Islam – chasing a group of people dressed in Islamic clothing, with the slogan "The Arab version of Godzilla."
Experts have observed the same pattern in other countries:
In the UK, after the knife attack in Southport, where a teenager killed three young women, AI was used to promote protests and post Islamophobic content based on false assumptions about the identity of the alleged attacker. An image depicted bearded men in traditional Islamic attire in front of the parliament, one waving a knife behind a crying child in a shirt with the British flag. And the message was: "We must defend our children!"
In France, AI-generated images were part of the far-right's campaign strategy for this year's European and legislative elections. Posts on X included images of people approaching a beach on boats with an anti-immigration slogan and videos criticizing Emmanuel Macron.
In Ireland, the use of AI content increased after the riots in Dublin in November 2023. A common image depicted Conor McGregor, the former MMA star who flirted with far-right rhetoric, standing in front of a bus in flames – a reproduction of an authentic image from the riots when a vehicle was set on fire in the city center. The AI-generated image received 20 million views.
And a campaign by the small far-right party People of Ireland included a post on X admitting to using an AI-generated image of a "supporter," saying he "barely recognizes" the place where he grew up.
In Italy, the populist party Lega published AI-generated images with messages against trans people and Islamophobia, including a "pregnant" bearded man, a group of men in Arab attire burning a copy of Dante's Divine Comedy, and one with Macron portrayed as a "EU soldier" in an anti-EU message. Political ads were posted on Facebook and Instagram and on Lega leader Matteo Salvini's social media accounts.
Other unmarked images from Salvini's account include pictures of women and children eating insects, referring to a popular right-wing conspiracy theory that the EU was preparing to force European citizens to eat insect-based foods.
You don't need to know programming to fake images
The use of the same models and themes and the similar approach in presenting images on social media platforms has given rise to a trend with an identical aesthetic in the EU and the UK. In all analyzed cases, far-right formations consistently avoided using any identifying markers on AI images, experts say.
"What we see is just the tip of the iceberg because what comes from individuals and beyond official channels is even worse," said Salvatore Romano, research head at the nonprofit organization AI Forensics, which examined posts from parties including the National Rally, Reconquest, and Les Patriots in France.
William Allchorn, senior researcher at Anglia Ruskin University, said that the ease with which AI models can be used has attracted a political fringe with a "pragmatic opportunism" towards new technologies.
"AI lowers the barriers to creating content. You don't need programming skills or anything similar to generate these images. It is also symptomatic that far-right views are being mainstreamed or normalized," he said, adding that the far right has no moral concerns about using AI images.
T.D.