Danish municipalities struggle with lengthy and error-filled AI-generated complaints

Danish municipalities are facing an increasing burden of overly long and inaccurate complaints from citizens using artificial intelligence tools to draft their grievances, according to a report by DR. Two AI experts now offer guidance on how to use the technology responsibly when submitting formal complaints.

Multiple municipalities report that complaints about administrative decisions have grown significantly longer, often containing factual errors or misinformation when generated by AI. While the technology can assist in formulating complaints, uncritical use risks overwhelming public caseworkers with poorly structured or incorrect submissions.

Prepare before generating text

Rikke Gade, associate professor at Aalborg University and AI expert, advises thorough preparation before using AI tools. “Do your groundwork so you know what the letter should contain,” she says. “What is the case about, and what legislation applies? You don’t need perfect phrasing—just outline the key points, and the tools can refine the language, both grammatically and stylistically.”

Peder Hammerskov, head of the Center for AI at the Danish School of Media and Journalism, recommends precision when instructing AI: “Specify that you need a 10-line response that clearly presents your point. Be as exact as possible in your prompt.”

Verify facts and limit sensitive data

Experts warn that AI-generated content may include “hallucinated” or outdated information, such as incorrect legal references. “It sounds convincing, but unless you’re an expert in the field, you should have someone with expertise review it—we can’t trust AI outputs blindly,” Gade explains.

Hammerskov suggests feeding the AI key terms from the municipal decision being contested to narrow the focus. He also cautions against including personal data, as AI models retain inputted information.

Review before submitting

Before sending a complaint, experts recommend:

• Reading the text carefully to ensure clarity and relevance
• Removing any AI-generated meta-commentary (e.g., “Here’s your draft”)
• Avoiding direct copy-paste to prevent the submission from appearing automated

“If it’s obviously AI-generated, it may be perceived as unserious,” Gade notes. Hammerskov encourages experimentation: “Try different approaches, refine gradually, and you’ll eventually produce something useful.”