Issue Bias in Generative AI Writing Assistance : Political Issues and LLMs in the Swedish 2026 Election
Bruinsma, Bastiaan ; Fredén, Annika LU
; Röttger, Paul
; Johansson, Moa
and Sayeed, Asad
(2026)
- Abstract
- Generative AI writing assistants and the Large Language Models (LLMs) that power them are increasingly part of how voters gather information before elections. With growing evidence that they influence users’ opinions, it is increasingly important to understand the views and positions of these tools. To better understand these views, we examine the stances supplied by six LLMs on a variety of Swedish-language writing tasks ahead of the 2026 Swedish parliamentary election. We cross 107 policy propositions with 77 writing templates and neutral, positive, and negative prompt framings, producing 24,717 prompts per model and 148,302 responses. To study these, we look at the models’ default stance tendencies, compare how they respond to similar... (More)
- Generative AI writing assistants and the Large Language Models (LLMs) that power them are increasingly part of how voters gather information before elections. With growing evidence that they influence users’ opinions, it is increasingly important to understand the views and positions of these tools. To better understand these views, we examine the stances supplied by six LLMs on a variety of Swedish-language writing tasks ahead of the 2026 Swedish parliamentary election. We cross 107 policy propositions with 77 writing templates and neutral, positive, and negative prompt framings, producing 24,717 prompts per model and 148,302 responses. To study these, we look at the models’ default stance tendencies, compare how they respond to similar issues, and compare their responses with those of each of Sweden’s eight parliamentary parties on the same issue. We find that Claude, DeepSeek, Gemini, and Mistral have similar profiles; ChatGPT more often supplies neutral or ambivalent text; and Grok differs most on topics such as migration, crime, and gender. When comparing the political parties, we find that the Social Democrats are closest to all six models. Still, after correcting for multiple comparisons, none of the within-model differences in party distances remains significant. Overall, we find that no model has a clear preference, nor a clear preference for a party, but that this depends on the specific issue or task the user asks about. (Less)
- Abstract (Swedish)
- Generative AI writing assistants and the Large Language Models (LLMs) that power them are increasingly part of how voters gather information before elections. With growing evidence that they influence users’ opinions, it is increasingly important to understand the views and positions of these tools. To better understand these views, we examine the stances supplied by six LLMs on a variety of Swedish-language writing tasks ahead of the 2026 Swedish parliamentary election. We cross 107 policy propositions with 77 writing templates and neutral, positive, and negative prompt framings, producing 24,717 prompts per model and 148,302 responses. To study these, we look at the models’ default stance tendencies, compare how they respond to similar... (More)
- Generative AI writing assistants and the Large Language Models (LLMs) that power them are increasingly part of how voters gather information before elections. With growing evidence that they influence users’ opinions, it is increasingly important to understand the views and positions of these tools. To better understand these views, we examine the stances supplied by six LLMs on a variety of Swedish-language writing tasks ahead of the 2026 Swedish parliamentary election. We cross 107 policy propositions with 77 writing templates and neutral, positive, and negative prompt framings, producing 24,717 prompts per model and 148,302 responses. To study these, we look at the models’ default stance tendencies, compare how they respond to similar issues, and compare their responses with those of each of Sweden’s eight parliamentary parties on the same issue. We find that Claude, DeepSeek, Gemini, and Mistral have similar profiles; ChatGPT more often supplies neutral or ambivalent text; and Grok differs most on topics such as migration, crime, and gender. When comparing the political parties, we find that the Social Democrats are closest to all six models. Still, after correcting for multiple comparisons, none of the within-model differences in party distances remains significant. Overall, we find that no model has a clear preference, nor a clear preference for a party, but that this depends on the specific issue or task the user asks about. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/a4a5f200-7933-4bb9-9490-3de8208dfee2
- author
- Bruinsma, Bastiaan
; Fredén, Annika
LU
; Röttger, Paul
; Johansson, Moa
and Sayeed, Asad
- organization
- alternative title
- Tendenser i hur generativ AI svarar politiskt : Politiska sakfrågor och stora språkmodeller i riksdagsvalet 2026
- publishing date
- 2026-09-28
- type
- Working paper/Preprint
- publication status
- published
- subject
- publisher
- arXiv.org
- project
- Bias and methods of AI technology studying political behavior
- language
- English
- LU publication?
- yes
- id
- a4a5f200-7933-4bb9-9490-3de8208dfee2
- alternative location
- https://arxiv.org/html/2609.15207
- date added to LUP
- 2026-09-22 13:56:11
- date last changed
- 2026-10-01 10:37:50
@misc{a4a5f200-7933-4bb9-9490-3de8208dfee2,
abstract = {{Generative AI writing assistants and the Large Language Models (LLMs) that power them are increasingly part of how voters gather information before elections. With growing evidence that they influence users’ opinions, it is increasingly important to understand the views and positions of these tools. To better understand these views, we examine the stances supplied by six LLMs on a variety of Swedish-language writing tasks ahead of the 2026 Swedish parliamentary election. We cross 107 policy propositions with 77 writing templates and neutral, positive, and negative prompt framings, producing 24,717 prompts per model and 148,302 responses. To study these, we look at the models’ default stance tendencies, compare how they respond to similar issues, and compare their responses with those of each of Sweden’s eight parliamentary parties on the same issue. We find that Claude, DeepSeek, Gemini, and Mistral have similar profiles; ChatGPT more often supplies neutral or ambivalent text; and Grok differs most on topics such as migration, crime, and gender. When comparing the political parties, we find that the Social Democrats are closest to all six models. Still, after correcting for multiple comparisons, none of the within-model differences in party distances remains significant. Overall, we find that no model has a clear preference, nor a clear preference for a party, but that this depends on the specific issue or task the user asks about.}},
author = {{Bruinsma, Bastiaan and Fredén, Annika and Röttger, Paul and Johansson, Moa and Sayeed, Asad}},
language = {{eng}},
month = {{09}},
note = {{Preprint}},
publisher = {{arXiv.org}},
title = {{Issue Bias in Generative AI Writing Assistance : Political Issues and LLMs in the Swedish 2026 Election}},
url = {{https://arxiv.org/html/2609.15207}},
year = {{2026}},
}