Despite AI’s promises of enhanced efficiency, communication and decision-making, there is an unsettling side to its use in politics. With growing concerns that AI could disrupt upcoming elections, it’s becoming clear that certain organisations aren’t employing the technology for the common good.
We unravel the darker sides of this alliance, exposing instances where political entities are tapping into the power of AI not as a force for progress, but rather as a tool for malevolent purposes.
AI-generated images
In March 2023, Norbert Kleinwächter, a right-wing German member of parliament from the AfD party, shared an image with a political message on Instagram. The distorted portrayal of refugees aimed to bolster the party’s anti-immigration stance, depicting Arab nationals as aggressive, shouting with exaggerated features and sharpened teeth. The scary AI-generated image bore the slogan “No to even more refugees!” Evidently, the sole purpose of this image was to cast the group in a negative light, perpetuating stereotypes and fostering increased animosity while garnering support for the party’s far-right policies.
While the image possesses cartoonish features, Kleinwächter has never disclosed that the picture was generated using artificial intelligence.
Numerous other instances reveal Kleinwaechter’s use of image-generating applications in crafting his Instagram political posts. Whether manipulating opposition photographs to fortify his political narrative or designing images to convey a specific message, he employs these tools with notable frequency. This spares the politician the effort of sourcing stock images and editing them with traditional digital platforms, like Photoshop. Utilising AI tools such as Midjourney, DALL-E or Adobe Firefly enables the creation of visually compelling scenes that precisely align with his political messaging.
Kleinwächter is a declared enthusiast of emerging technologies, actively seizing opportunities presented by them. While embracing the potential of software applications, he strongly opposes regulation of artificial intelligence and the digital space. Examining his online activities to what ends the politician harnesses these technological advances.
AI-generated campaign posters
Another example of AI-generated campaign posters comes from the FDP Schweiz, the Swiss liberal party, which employed artificial intelligence imagery in their campaign against the climate activist group Renovate Switzerland. The group uses tactics akin to those of Letzte Generation in Germany and Extinction Rebellion in the UK, including creating roadblocks by glueing themselves to the road to dramatise warnings of an impending climate disaster.
The generated image looks convincing. However, it is actually an AI-generated attempt to denounce the activists’ actions by highlighting the disruption caused, artificially depicting an ambulance unable to pass through the human chain. While this scenario is plausible, there is no documented evidence that such an event actually occurred or was captured in a real photograph.
AI-generated video
Moving to dynamic content, Jimmie Åkesson, leader of the hard-right Sweden Democrats, used artificial intelligence to deliver a populist speech in Arabic telling immigrants and asylum seekers they are not welcome and should go home.
While this may be perceived as an innovative means of connecting with different linguistic groups in society, critics contend that the address was primarily directed at his Swedish-speaking supporters, not at Arabic-speakers.
Similarly, Loic Signor, the spokesperson for French President Emmanuel Macron’s centrist Renaissance party, posted a video on Twitter purportedly featuring Marine Le Pen delivering her New Year’s message in both French and Russian. Clearly intending to suggest the far-right French politician’s alignment with Russia, the AI-generated video disseminated misinformation as a strategy to counter opposition.
In a similar confrontational manoeuvre, the radical left-wing activist group Zentrum für Politische Schönheit released a fabricated AI video in November featuring Chancellor Olaf Scholz denouncing the far-right AfD. This action was part of a broader political campaign seeking to ban the party due to its extremist policies.
Sometimes reality catches up with AI-generated fiction. Recently, an investigative report by Correctiv exposed a clandestine meeting among far-right leaders discussing plans for a large-scale “remigration” of both refugees and German citizens of immigrant origin. The revelations sparked widespread protests, with calls for banning the party due to its resemblance to Nazi ideology.
In response, Scholz expressed support for the protests against the far-right, even participating in the demonstrations himself.
Despite the chancellor’s stance, the impact of the original deep fake remains uncertain; and its ethical implications are questionable.
Ethics of AI in politics
These examples illustrate abuses of AI by political actors to construct convincing yet inaccurate depictions of themselves or their adversaries.
The portrayal of immigrants exclusively as loud and disruptive, as promoted by the AfD, deliberately distorts reality. Similarly, the visual narrative of climate activists obstructing an ambulance through roadblocks is not supported by evidence.
While AI holds potential for enhancing political communication, especially in addressing diverse linguistic groups through translation, recent examples underscore its misuse for dubious ends.
The strategy of fighting fire with fire, or combating misinformation with misinformation, raises ethical concerns and can undermine the credibility of the political actors involved, especially when such videos are not clearly labelled as fake from their inception.
These images undeniably draw considerable attention—a crucial goal in political campaigns. However, using AI technology to stretch and distort reality in the interests of a political organisation is a danger to democracy and to freedom of thought.