Artificial intelligence is becoming part of the political writing process. According to a recent study, politicians in the United Kingdom and Sweden may be using AI to draft parliamentary speeches, statements and motions without disclosing that assistance to the public.

Researchers from Chalmers University of Technology in Sweden and Edinburgh Napier University in the UK analysed 13,565 Swedish parliamentary motions and 4,209 British parliamentary statements published between 2021 and April 2026.

Their findings suggest that the use of generative AI in parliamentary communication has increased rapidly in both countries.

AI-assisted political texts are becoming more common

In 2026, approximately 15% of the analysed UK parliamentary statements contained at least one paragraph classified as AI-generated.

In Sweden, 9.4% of the motions submitted during the 2025–2026 parliamentary year contained at least one paragraph identified as potentially AI-assisted. This represents around 300 motions.

The researchers found no disclosure of AI use in any of the texts flagged by their detection systems.

The increase is significant. In the UK, the estimated proportion was approximately 3.5% in 2022. In Sweden, it stood at 1.7% during the 2022–2023 parliamentary period.

When the researchers analysed complete documents rather than individual passages, 6.5% of Swedish motions from the latest parliamentary year were classified as AI-generated. In the UK, the equivalent figure for 2026 was 2.1%.

Why transparency matters

Using AI to research a subject, improve clarity or review a draft is different from asking a model to produce an entire parliamentary proposal. However, without disclosure, voters cannot know how much human judgement went into the final document.

Minerva Suvanto, the study’s lead author and a PhD student at Chalmers University of Technology, warned that the lack of transparency creates risks regardless of how extensively AI was used.

AI-generated political writing can sound polished, confident and persuasive even when it contains inaccurate or misleading information. This creates the possibility that errors will be overlooked not only by readers, but also by the politicians or staff members publishing the material.

The issue is therefore about more than authorship. It concerns accountability for the arguments, facts and policy decisions presented in official parliamentary documents.

How the researchers identified potential AI use

The research team developed separate detection systems for Swedish and English parliamentary texts.

The systems were trained using human-written documents created before generative AI became widely available, together with AI-generated texts covering similar political topics. The detectors were then applied to parliamentary material published from 2021 onwards.

Researchers examined both complete documents and individual paragraphs. A document was classified as containing AI-assisted content when at least one sufficiently long paragraph was identified as likely AI-generated.

To evaluate the systems, the researchers tested them on parliamentary texts from 2021, before ChatGPT became publicly available. Only one complete document from each country was incorrectly classified as AI-generated.

The researchers describe their method as “interpretable AI”, meaning that the system can indicate which passages influenced its classification. It does not, however, identify individual “AI words”.

As with other AI-detection methods, the system has important limitations. The study cannot determine which model was used, how the politician interacted with it or whether AI generated an entire passage, rewrote a human draft or simply improved its style.

Politicians are already experimenting with AI

Some political leaders have publicly acknowledged using generative AI as part of their work.

In August 2025, Swedish Prime Minister Ulf Kristersson said he regularly used tools such as ChatGPT and the French chatbot Le Chat to obtain a “second opinion” on political questions. According to his press secretary, no security-sensitive information was entered into these systems.

Jan Riise, a Swedish Green Party MP and spokesperson on AI and digitalisation, said his party uses AI for research and has an internal policy encouraging caution. He also argued that politicians should disclose when AI has played a substantial role in producing a document.

These examples show that AI use in politics is no longer hypothetical. The more important questions now concern where institutions draw the line and how they communicate that line to the public.

From productivity tool to governance challenge

Generative AI can help political teams research complex subjects, summarise information and prepare documents more efficiently. Yet parliamentary texts are not ordinary marketing or administrative materials. They can influence public debate, shape legislation and affect the lives of millions of people.

This makes human oversight essential. Every claim must be verified, every recommendation must be reviewed and a clearly identified person must remain accountable for the final text.

Public institutions may therefore need explicit rules covering acceptable use, disclosure, fact-checking, sensitive information and responsibility for AI-assisted content.

The central issue is not whether politicians should be allowed to use AI. It is whether citizens deserve to know when technology has helped produce the words, arguments and proposals presented in their name.

Source