In Germany's 2024 elections (EU Parliament and subnational elections of Thuringia, Saxony, and Saxony Anhalt) AI served primarily as a tool for narrative reinforcement or propaganda rather than direct attacks, with the AfD leading in strategic deployment of AI-generated content and algorithmic manipulation—while smaller parties leveraged AI to overcome resource constraints, highlighting both democratizing potential and risks to media trust and political cohesion.
Share
The year 2024 marks a pivotal moment in digital political communication in Germany: for the first time, artificial intelligence (AI) is playing a decisive role in election campaigns. This development offers opportunities to foster democratic innovation and citizen participation but also poses significant risks, such as disinformation and the fragmentation of political discourse.
This study, available in German, provides an in-depth analysis of the dynamics and implications of AI in the political sphere, with a particular focus on the European elections and state elections.
KI und Wahlen: Neue Dynamiken der digitalen Demokratie in Deutschland
AI-Incidents
In Germany, two major elections occurred during the study period. The first was the European elections, held from June 6 to June 9, 2024. The second involved three regional elections: the elections in Thüringen and Sachsen took place on September 1, 2024, while the regional election in Brandenburg occurred on September 22, 2024.The analysis of AI-related incidents spanned the period from January 1, 2024, to September 23, 2024. A total of 4,140 news articles were analyzed, resulting in the identification of 37 AI-related cases.
Trends:
- Reinforcing narratives through AI-generated content
-
The use of artificial intelligence in the EU and state elections was not primarily aimed at discrediting political opponents, but rather at strengthening its own narrative.
- Regular use of AI-generated imagery by the AfD
-
The AfD deliberately uses AII-generated content, such as deepfakes, to transform complex political issues into simple, visual messages. The images are often not labeled as AI-generated.
- Enabling cheaper political communication
-
AI can help to develop a coherent branding for politicians and parties. This includes not only text-based content, but also AI-generated music that can be used to underscore campaigns and appeal to voters. This saves resources and content can be created quickly and flexibly.
- Targeted use of the workings of social media algorithms
-
The AfD has effectively utilized social media platforms to strategically disseminate its content. The party predominantly leverages emotional appeal and promotes interaction with its content by collectively mobilizing its supporters. Social media algorithms prioritize interactive content, enhancing its visibility and allowing it to reach a broad audience on platforms such as TikTok, even without the need for political advertising.
- Electoral success through data-driven campaign management
-
By analyzing voter behavior and demographic trends, key constituencies can be identified and campaigns targeted.
- Targeted political advertising on online platforms
-
Political parties are increasingly turning to online advertising on platforms such as Meta and Google to show their digital presence and reach their target groups.
- Strengthening smaller parties through AI
-
As the BSW and Volt have shown in the EU and national elections, artificial intelligence can help level the playing field for smaller parties compared to large, established parties. AI's many applications can help smaller parties overcome resource constraints, for example by automating tasks that would otherwise require human or cost-intensive resources.
- Accessibility of political campaigns
-
More and more parties are creatively using AI to make their digital election campaigns more personal and accessible with chatbots. In addition, large language models (LLMs) are being used as sparring partners to develop different perspectives on issues and prepare for political debates.
Proxy Variables:
- Trust in News
- Source of News
- Social Media Usage
- Trust in Media and Government
Vulnerabilities
- Trust in media is at stake
-
Advancing AI development is increasingly overcoming the "uncanny valley effect" - the phenomenon in which artificial representations are perceived as recognizable scary (Breuer, 2024). When this natural protection mechanism disappears, people lose the intuitive ability to distinguish between real and AI-generated content. This can fundamentally undermine trust in media content and thus the basis for public opinion formation.
- Political fragmentation through hyper-personalization
-
Social media and AI promote the hyper-personalization of political speech, which means that political information is increasingly tailored to the views of users. This can reinforce one's own political beliefs and lead to the exclusion of other opinions. This promotes political fragmentation, which can lead to increasing polarization.
- Collective Mobilization and Platform Competence
-
At the heart of every platform is its algorithm, which determines whether content on the platform is given reach. The AfD skillfully uses its digital infrastructure (Nicolaus & Scherndel, 2024) to achieve maximum online presence on various platforms. Most of the time, it is emotional, simple content that generates a lot of engagement. This strategy, which skillfully exploits "algorithmic competence," is used by various actors worldwide to generate visibility and reach and is highly effective in supporting collective mobilization on and offline (Munoz, 2024).
- Liar's Dividend
-
The more realistic deepfakes become, the easier it is for authentic statements to be dismissed as AI-generated.
- Platform policies for AI-generated content are not consistently enforced
-
Most platforms have policies regarding the use of AI-generated imagery that require labeling. However, platforms are not consistently enforcing these policies.
