AI in elections: who sets the rules? | DW News

Key Concepts

• AI-Generated Political Content: The use of artificial intelligence to create images, audio, or video for campaign purposes.
• Transparency/Disclosure: The practice of informing the public when content is AI-generated.
• Deepfakes/Synthetic Media: AI-manipulated content that mimics real people or events.
• Regulatory Vacuum: The current lack of specific legal frameworks governing AI use in political campaigning.
• EU AI Act: Upcoming legislation requiring mandatory labeling of AI-generated content.

The State of AI in German Election Campaigns

Political parties in Germany are increasingly integrating AI into their campaign strategies. According to David Fischer, a political campaign advisor and researcher, AI is being utilized across the political spectrum, yet it is frequently deployed without public disclosure. Fischer’s survey of parties in German states holding elections this year revealed that a majority of parties failed to respond to inquiries regarding their AI usage, signaling a systemic disregard for transparency.

Real-World Applications and Misuse

The video highlights several instances where AI has been used to influence voter perception:

• Promotional Content: A Green Party politician used AI to create a video of a bike loop to promote infrastructure, failing to disclose the AI origin.
• Satirical/Political Messaging: A CDU (Conservative) politician produced an AI-generated rap song about the city budget, with disclosure buried only in the caption.
• Deceptive Content: The AfD (Far-right) party used AI to put fake words into the mouth of a state minister. The disclosure was only added after media outlets exposed the manipulation.
• Negative Campaigning: A regional CDU branch used AI to depict a Minister of Education surrounded by piles of cash to criticize school funding, raising concerns about whether voters will retain the "AI-generated" label once the emotional impact of the image takes hold.

The "Wild West" of Campaigning

The current political landscape is characterized by a lack of standardized rules. Research from the "Campaign Tracker" project during the previous federal election identified approximately 1,000 AI-generated posts, with only 13% being properly labeled. This low compliance rate suggests that political parties are currently self-regulating, often prioritizing impact over honesty.

Regulatory Outlook

The primary challenge identified is the ambiguity regarding what constitutes "misleading" content and who holds the authority to define it. Currently, there is no consensus among German political parties to establish shared, voluntary rules for AI usage.

However, legislative change is on the horizon:

• EU AI Act: Starting in August 2026, European Union law will mandate that all AI-generated content on social media must be clearly labeled.

Synthesis and Conclusion

The core issue is not the existence of AI tools, but the "silent" nature of their deployment. By failing to disclose AI usage, political parties are eroding voter trust and transparency. While the EU AI Act will eventually provide a legal framework for mandatory labeling, the current environment remains largely unregulated, allowing parties to manipulate public perception through synthetic media without accountability. The takeaway is that until clear, enforceable standards are adopted, voters must remain highly skeptical of digital political content.