AI is playing an increasingly visible role in political communication, including at the local level. Researchers at the University of Amsterdam are using the new CampAIgn Tracker to monitor how political parties and candidates use AI in the 2026 Dutch municipal elections. The first findings show that AI-generated campaign imagery is widespread, while transparency is largely absent.
Researchers at the University of Amsterdam identified 192 posts containing AI-generated images in the first ten days of monitoring, which began on 1 March. The content ranges from realistic deepfakes to stylised graphics and was found across Facebook, Instagram, TikTok and X. According to political communication researcher Fabio Votta, who coordinates the project, the PVV is once again the largest producer, just as it was during the national elections.
The most striking finding is that around 90 percent of this AI-generated campaign content was published without any label or disclosure. This matters because generative AI models are becoming increasingly capable of producing highly realistic images, making it harder for voters to distinguish authentic content from synthetic material. Transparency is therefore crucial, especially with the EU AI Act set to fully enter into force in August, requiring AI-generated content to be clearly labelled.
The most striking finding is that around 90 percent of this AI-generated campaign content was published without any label or disclosure. This matters because generative AI models are becoming increasingly capable of producing highly realistic images, making it harder for voters to distinguish authentic content from synthetic material. Transparency is therefore crucial, especially with the EU AI Act set to fully enter into force in August, requiring AI-generated content to be clearly labelled.
One of the more notable patterns is that AI use appears across almost all municipalities being tracked, but some cities stand out. Lelystad currently accounts for 46 AI items, of which 37 come from PVV candidates. Steenbergen follows with 12 AI items, all published by one local party, the Volkspartij, which researchers describe as posting so-called AI slop-content.
Alongside the tracker, the researchers also launched the interactive tool “AI or Not?”, allowing users to test whether they can identify AI-generated campaign images. The project is part of the Hybrid Election Integrity Observatory (HEIO)consortium, together with Post-X Society, Trollrensics and AI Forensics.
Read also
