The case against AI in Politics
25/9/26
AI is inevitable - I accept that.
I accept that discovery, innovation and industrialisation has been the cornerstone of what has made my country great, and equally I accept that AI is a new chance for discovery.
So why is AI becoming a part of political processes so hard to accept?
According to this study, up to 15% of all Parliamentary texts in the UK and Sweden were AI-generated. However, this has long-been public knowledge in the UK. As a matter of fact, the phrase “I rise to speak” - a phrase common over 200 years ago - was used over 600 times in the House of Commons in 2025. Having not been used in hundreds of years, how did it once again find home in British Politics? Well, where it is commonly used, however, is in the United States. And the most common LLM tools to write speeches - are American AI companies.
Speeches are not the standout case here, however. Reports, briefings and papers are all repeatedly failing AI-detectors, as seen in this Economist article. Where this does get interesting, is the fact that AI-detectors do not detect how AI has been used - only that it has.
So, if one were to double-check translations or spellings through AI, and the AI model changes one small error - that text will fail an AI detector. As Godfrey argues in the Economist, once the public sees that a text has failed an AI detector, it is too late to explain how it was used, as the fact that it was, is enough to damage an already lacking trust.
Social Media has created a political landscape where voters have unrestricted access to their elected officials. What they eat, drink, wear, do, date - is all public knowledge, and is how trust between elected officials and the electorate in the 21st century is built, whether they like it or not. Hereby lies the issue, no matter how far technology advances, human nature remains. The human desire for connection, relatability and authenticity exceeds technological innovation.
So, when the people who we entrust to lead our countries show themselves to be as shallow as a skin of robot parts, voters lose trust. Not just in the people they voted for, but for the institutions that allow this to continue.
But this writing is not to argue against AI. Instead, it is to encourage human connection where it is needed most.
We advocate for causes closest to our heart, not our processor.
We inspire people by making them think, not generate responses.
We connect with people by sharing experiences, not by automated answers.
And that has been the case throughout every dawn of technological innovation, and I see the 21st Century AI-revolution as no different.
Because we are human, and so human instinct will always win.
- Godfrey, C. (2026) 'Is AI writing taking over Westminster?', The Economist, 9 July. Available at: https://www.economist.com/britain/2026/07/09/is-ai-writing-taking-over-westminster
- Suvanto, M. et al. (2026) Detecting undisclosed LLM-generated content in parliamentary texts. arXiv:2606.14209. Available at: https://arxiv.org/abs/2606.14209
