Uncovering Political Bias in Large Language Models using Parliamentary Voting Records

2026-01-13

AI Summary

This study addresses the lack of systematic evaluation of political bias in large language models (LLMs) grounded in real-world political behavior, particularly the absence of cross-national comparable benchmarks. The authors propose the first political bias assessment framework that integrates parliamentary voting records, expert survey data from the Chapel Hill Expert Survey (CHES) for ideological mapping, and natural language analysis. They construct a multilingual evaluation dataset covering the Netherlands, Norway, and Spain. By aligning model-generated political positions with actual party voting behavior and leveraging the CHES two-dimensional ideological space for interpretable cross-party comparisons, the study reveals that mainstream LLMs consistently exhibit left-leaning or centrist tendencies and display significant negative bias toward right-wing conservative parties, thereby demonstrating the framework’s effectiveness and novelty in uncovering fine-grained political orientations.

Technology Category

Machine Learning: Large Multimodal Models (LMMs)Natural Language Processing: (Large) Language ModelsPhilosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendationGraph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphsWeb Mining and Content Analysis: Large pretrained models with web data

Abstract

As large language models (LLMs) become deeply embedded in digital platforms and decision-making systems, concerns about their political biases have grown. While substantial work has examined social biases such as gender and race, systematic studies of political bias remain limited, despite their direct societal impact. This paper introduces a general methodology for constructing political bias benchmarks by aligning model-generated voting predictions with verified parliamentary voting records. We instantiate this methodology in three national case studies: PoliBiasNL (2,701 Dutch parliamentary motions and votes from 15 political parties), PoliBiasNO (10,584 motions and votes from 9 Norwegian parties), and PoliBiasES (2,480 motions and votes from 10 Spanish parties). Across these benchmarks, we assess ideological tendencies and political entity bias in LLM behavior. As part of our evaluation framework, we also propose a method to visualize the ideology of LLMs and political parties in a shared two-dimensional CHES (Chapel Hill Expert Survey) space by linking their voting-based positions to the CHES dimensions, enabling direct and interpretable comparisons between models and real-world political actors. Our experiments reveal fine-grained ideological distinctions: state-of-the-art LLMs consistently display left-leaning or centrist tendencies, alongside clear negative biases toward right-conservative parties. These findings highlight the value of transparent, cross-national evaluation grounded in real parliamentary behavior for understanding and auditing political bias in modern LLMs.

Problem

• political bias
• large language models
• parliamentary voting records
• ideological bias
• LLM evaluation

Innovation

• political bias
• parliamentary voting records
• large language models
• CHES ideology space
• bias benchmarking