Assessing Human Rights Risks in AI: A Framework for Model Evaluation

2025-10-06

AI Summary

This paper addresses the potential threats posed by generative AI to fundamental human rights—including non-discrimination, health, and safety—by proposing the first computational risk assessment framework systematically integrating the United Nations’ Guiding Principles on Business and Human Rights (UNGPs). Methodologically, it establishes a three-stage, actionable workflow: task selection, risk indicator design, and rights impact analysis—leveraging algorithmic auditing and NLP techniques to quantify AI’s real-world effects on rights such as access to information and freedom of thought in concrete scenarios (e.g., political news generation). Its key contribution lies in the first structured incorporation of international human rights standards into AI evaluation, enabling cross-model comparison and benchmarking. Empirical evaluation on large language models demonstrates the framework’s capacity to reliably differentiate human rights risks across models, offering a reproducible, scenario-driven paradigm for AI ethics assessment.

Technology Category

• Philosophy and Ethics of AI: Bias, Fairness & Equity
• Humans and AI: AI for Accessibility
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

The Universal Declaration of Human Rights and other international agreements outline numerous inalienable rights that apply across geopolitical boundaries. As generative AI becomes increasingly prevalent, it poses risks to human rights such as non-discrimination, health, and security, which are also central concerns for AI researchers focused on fairness and safety. We contribute to the field of algorithmic auditing by presenting a framework to computationally assess human rights risk. Drawing on the UN Guiding Principles on Business and Human Rights, we develop an approach to evaluating a model to make grounded claims about the level of risk a model poses to particular human rights. Our framework consists of three parts: selecting tasks that are likely to pose human rights risks within a given context, designing metrics to measure the scope, scale, and likelihood of potential risks from that task, and analyzing rights with respect to the values of those metrics. Because a human rights approach centers on real-world harms, it requires evaluating AI systems in the specific contexts in which they are deployed. We present a case study of large language models in political news journalism, demonstrating how our framework helps to design an evaluation and benchmarking different models. We then discuss the implications of the results for the rights of access to information and freedom of thought and broader considerations for adopting this approach.

Problem

• Developing computational framework to assess AI human rights risks
• Evaluating model risks to non-discrimination health security rights
• Measuring potential harms in specific AI deployment contexts

Innovation

• Framework computationally assesses human rights risk
• Approach evaluates models for specific human rights
• Metrics measure scope scale likelihood of risks