Researchers Detect Increasing Use of AI in Parliamentary Motions Without Disclosure
Research from Chalmers University of Technology reveals a growing trend of politicians using AI to draft parliamentary motions without open disclosure. AI detectors were developed to identify this usage.
A joint study by Chalmers University of Technology and Edinburgh Napier University has developed sophisticated AI detectors capable of identifying the use of artificial intelligence in parliamentary documents. The research highlights a significant increase in AI-assisted content within political motions in both Sweden and the UK.
The study analysed thousands of parliamentary texts and found that while AI usage is on the rise, a notable absence of disclosure from politicians is concerning. Researchers point out potential risks, including the persuasion of AI-generated text and the possibility of unspotted errors, which could undermine public trust.
The developed AI detection tools are based on machine learning trained on extensive datasets of Swedish and British parliamentary records. Employing "interpretable AI," the system not only flags AI-generated content but also pinpoints specific sections within the text where AI assistance is evident.
Data indicated a sharp rise in AI-assisted texts in Swedish parliamentary documents, growing from near zero in 2021–2022 to approximately 10% by the 2025–2026 parliamentary year. Similar trends were observed in the UK, with no significant differences noted across political parties or countries.
This research underscores the need for transparency in the political sphere concerning AI adoption, especially following the implementation of the EU's AI Act. The findings suggest that political actors may need to reconsider their reporting of AI use to maintain voter confidence.