Representational Harms in LLM-Generated Narratives Against Global Majority Nationalities

2026-04-24

AI Summary

This study addresses systematic representational harms inflicted by large language models (LLMs) on the national identities of global majority populations in narrative generation, manifesting as stereotyping, identity erasure, and role homogenization. Through open-ended prompting combined with prompt engineering, narrative content analysis, and cross-national controlled experiments, this work provides the first evidence—from the perspective of global majority groups—that LLMs persistently exhibit U.S.-centric bias even when nationality prompts are explicitly substituted, thereby challenging the “model sycophancy” hypothesis. The findings reveal that characters with minority national identities appear in subordinate roles over 50 times more frequently than in leading roles, and that cues signaling U.S. nationality significantly amplify such biases, underscoring the need for caution in deploying U.S.-centric LLMs in global contexts.

Technology Category

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

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Large language models (LLMs) are increasingly used for text generation tasks from everyday use to high-stakes enterprise and government applications, including simulated interviews with asylum seekers. While many works highlight the new potential applications of LLMs, there are risks of LLMs encoding and perpetuating harmful biases about non-dominant communities across the globe. To better evaluate and mitigate such harms, more research examining how LLMs portray diverse individuals is needed. In this work, we study how national origin identities are portrayed by widely-adopted LLMs in response to open-ended narrative generation prompts. Our findings demonstrate the presence of persistent representational harms by national origin, including harmful stereotypes, erasure, and one-dimensional portrayals of Global Majority identities. Minoritized national identities are simultaneously underrepresented in power-neutral stories and overrepresented in subordinated character portrayals, which are over fifty times more likely to appear than dominant portrayals. The degree of harm is amplified when US nationality cues (e.g., ``American'') are present in input prompts. Notably, we find that the harms we identify cannot be explained away via sycophancy, as US-centric biases persist even when replacing US nationality cues with non-US national identities in the prompts. Based on our findings, we call for further exploration of cultural harms in LLMs through methodologies that center Global Majority perspectives and challenge the uncritical adoption of US-based LLMs for the classification, surveillance, and misrepresentation of the majority of our planet.

Problem

• Representational Harms
• Large Language Models
• Global Majority
• National Origin Bias
• Stereotyping

Innovation

• representational harms
• Global Majority
• LLM bias
• narrative generation
• cultural bias