How the AfD on the network at AI depends on the competition

A glowing hot summer day. A young woman is alone in the swimming pool. Her eyes are torn open, the view flicker. Behind her: a group of dark -haired men who stare at them to finally storm towards them. One voice warns from off: “The number of sexual attacks in Bavaria’s swimming pools rose by 18 percent in 2024 compared to the previous year. Almost 60 percent of the suspects are foreigners.”

The pictures are not real. The woman does not exist. The men are also mere pixels. Everything about this video is done by artificial intelligence (AI). But it looks like real. It is a so -called Deeppake.

The video comes from the AfD and is spread wherever the party is traditionally represented: on Tikok, Facebook, X and other channels on the Internet.

There are many other such examples. No other parliamentary party works equally as unscrupulously and effectively with a generative AI as the AfD. An evaluation of the University of Mainz of over 70,000 social media posts for the Bundestag election showed that only 500 advertising contributions (0.3 percent) contained AI elements. But with the AfD, the proportion was ten times higher (over three percent). And that should only be the beginning.

A look at the statistics shows that social media can be crucial for the range and the effect of political messages today: At Tikok, the AfD comes together to more than 620,000 followers – more than the CDU, SPD and the Greens.

At almost 410,000 only, the left is not quite as ago. Party leader Alice Weidel has almost a million followers-significantly more than left-wing politician Heidi Reichinnek (620,000). The numbers on Instagram look a little different – there Heidi Reichinnek with 766,000 followers of Alice Weidel briefly surpasses (750,000 followers).

Social media are the natural habitat of the AfD. In contrast to the established media, she does not have to deal with research or put up with criticism. Instead, your political business model works particularly well.

“The AfD has little fear of contact with new platforms,” says political and digital consultant Martin Fuchs. “She understood what gold is for the algorithms of the entertainment platforms, namely: polarization, escalation, provocation and emotions.”

This is particularly visible on Tiktok, says Fuchs. For four years, the AfD had almost propagated there before the other parties were followed up. “These are four years of experience.” During this time, the party built a digital apron, from supporters who made advertising on their own. “Eight out of ten contributions about the AfD do not even come from the party itself,” says Fuchs. That means “Organic Content”.

And now the AI is added, where the party has an advantage again. “Of course, the AfD does not play with fair cards here,” says the political advisor. “But it moves in the gray area of what is still okay by law.”

What Fuchs says: The party and its supporters do not characterize the articles generated with AI. You thus ignore the EU’s AI regulation (AI Act), which regulates clearly: Whoever publishes Deepfakes must make it clear that they are artificially generated. Exceptions only apply to obviously satirical and creative works.

However, the regulation part, which prescribes the label, only formally comes into force in one year. Accordingly, the AfD gives itself to a request from the star. The effects of the EU regulation are currently being “examined,” said a party spokesman. You use “both common providers and in-house AI programs”.

From the perspective of AI legal expert Andreas Daum, however, the AfD takes a risk in the medium term. The party is still only classified as an operator. However, if you are strongly adapting AI software for your own purposes, you can be considered a provider of a “high-risk AI system”. Then they suspect stricter regulations.

As a precaution, the AfD refers to “internal test processes” and “mandatory training”. But that sounds little convincing for Daum. “Those responsible must ensure that the AI users have a sufficient understanding of AI-including social and ethical implications,” he says. “And you also have to train your staff in which cases there is an inadmissible AI mission in the election campaign.”

For Martin Fuchs, the technical question is not even the biggest problem. “The content that is shown is much more decisive,” he says. AI makes it possible to create deceptively real pictures for every imaginable scenario. As an example, Fuchs mentions alleged “knife men” who are currently committing an act of violence. “There are no such pictures in any database,” says Fuchs. “So you let them generate” and create “felt realities”.

According to Fuchs, AI will play a much larger role in the next election campaigns. What is missing is a real debate in political space: “How do we want to use AI? What is AI? Where is our red line? What is going – and what not?” There are initial steps, for example at the Left Party or the former FDP faction that had developed internal code of conduct. But that is not enough.

Overall, awareness of the problem is still too low, confirms SPD digital politician Johannes Schätzl. The digital policy spokeswoman for the left faction, Donata Vogtschmidt, in turn accuses the AfD of consciously taking advantage of existing regulatory gaps. Therefore, sharper controls and structural reforms are necessary, she says. Instead of platforms that are designed for range and advertising, common good -oriented alternatives are needed.

But reality is still different. And even in her, Martin Fuchs does not see any automatism that speaks for the AfD: “For the last Bundestag election, the SPD and Left on Tiktok were even more successful.”