Political Ideology Shifts in Large Language Models

2025-08-21

AI Summary

This study investigates the ideological plasticity of large language models (LLMs) when prompted with synthetic personas. Using the Political Compass test, we systematically evaluate the political orientation responses of seven LLMs—spanning diverse parameter scales—to multiple role-based prompts, including right-authoritarian and left-libertarian personas. Our findings reveal: (1) scaling up model size significantly expands the implicit ideological spectrum and intensifies polarization, particularly under right-authoritarian prompts; (2) persona descriptions induce systematic, directional ideological shifts; and (3) ideological expression is highly context-dependent and malleable via prompt engineering. Crucially, this work provides the first empirical evidence of a synergistic interaction between model scale and persona prompting. It establishes a foundational framework for modeling LLM political behavior, advancing safety alignment strategies, and informing transparency-aware governance in high-stakes deployment scenarios.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Humans and AI: Learning Human Values and Preferences

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Large language models (LLMs) are increasingly deployed in politically sensitive settings, raising concerns about their potential to encode, amplify, or be steered toward specific ideologies. We investigate how adopting synthetic personas influences ideological expression in LLMs across seven models (7B-70B+ parameters) from multiple families, using the Political Compass Test as a standardized probe. Our analysis reveals four consistent patterns: (i) larger models display broader and more polarized implicit ideological coverage; (ii) susceptibility to explicit ideological cues grows with scale; (iii) models respond more strongly to right-authoritarian than to left-libertarian priming; and (iv) thematic content in persona descriptions induces systematic and predictable ideological shifts, which amplify with size. These findings indicate that both scale and persona content shape LLM political behavior. As such systems enter decision-making, educational, and policy contexts, their latent ideological malleability demands attention to safeguard fairness, transparency, and safety.

Problem

• Investigating ideological expression shifts in LLMs under synthetic personas
• Assessing scale-related susceptibility to explicit political cues
• Examining systematic ideological shifts induced by persona thematic content

Innovation

• Using synthetic personas to probe ideological expression
• Leveraging Political Compass Test as standardized measurement
• Analyzing scale-dependent susceptibility to ideological priming