Estimating the Geopolitical Preferences of Large Language Models from United Nations Voting Data

2026-07-28

AI Summary

Existing approaches struggle to accurately characterize the complex geopolitical preferences of large language models (LLMs). This study pioneers the systematic application of the dynamic ordinal ideal point model—a well-established framework in international relations—to LLMs, treating them as respondents to 5,555 United Nations General Assembly resolutions from 1946 to 2025. By inferring stance positions from each model’s expressed support for the full texts of these resolutions, the method enables large-scale quantitative analysis of LLMs’ geopolitical orientations. The findings reveal that LLMs frequently diverge significantly from the official positions of their countries of origin: GPT-5, Claude Sonnet, and Gemini align more closely with Russia in the 21st century, while DeepSeek gravitates toward France. Notably, on issues involving U.S.–China–Russia rivalries, support for Chinese and Russian positions reaches as high as 96.1% across models.

Abstract

How should researchers measure the geopolitical preferences expressed by large language models (LLMs)? Existing audits commonly rely on surveys and simple tests, but international-relations research has long recognized that measuring geopolitical preferences is difficult and has developed methods for recovering them from observed choices. This paper applies a dynamic ordinal ideal-point approach from international relations, treating LLMs as respondents to the full texts of 5,555 divisive, recorded, adopted resolutions considered in regular sessions of the UN General Assembly from 1946 through 2025. Support ranges from 37.8% for DeepSeek to 97.3% for GPT-5. Surprisingly, in the twenty-first century, GPT-5, Claude Sonnet, and Gemini are closest among the permanent five to Russia; DeepSeek is closest to France; and all four are farthest from the United States. Among 2,104 resolutions opposed by the United States but supported by China and Russia/USSR, GPT-5 supported 96.1%, Gemini 83.4%, Claude Sonnet 65.2%, and DeepSeek 36.1%. The findings show that a model's expressed geopolitical position can differ markedly from that of its developer's home country, especially in international politics, where state actions can diverge from the stated principles prevalent in the texts on which models are trained.

Problem

• geopolitical preferences
• large language models
• United Nations voting
• ideal-point estimation
• international relations

Innovation

• dynamic ordinal ideal-point
• geopolitical preference estimation
• large language models
• UN General Assembly voting
• AI alignment auditing