Assessing Model-Agnostic XAI Methods against EU AI Act Explainability Requirements

2026-03-19

AI Summary

This study addresses the gap between model-agnostic explainable artificial intelligence (XAI) methods and the interpretability requirements stipulated in the European Union’s Artificial Intelligence Act, which has hindered effective regulatory compliance. To bridge this disconnect, the work proposes a structured mapping mechanism that aligns XAI technical attributes with specific legal provisions of the Act. Integrating qualitative expert assessments, multidimensional analysis of XAI characteristics, and detailed regulatory interpretation, the authors develop a quantifiable compliance scoring framework. This framework systematically evaluates the extent to which existing XAI approaches fulfill legal obligations for explainability, while also identifying current technological limitations and ambiguities in regulatory language. The resulting tool offers both theoretical insights and practical guidance to support compliance efforts, inform technical refinements, and contribute to future policy development.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Philosophy and Ethics of AI: Accountability, Interpretability & Explainability
• Machine Learning: Transparent, Interpretable, Explainable ML

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Explainable AI (XAI) has evolved in response to expectations and regulations, such as the EU AI Act, which introduces regulatory requirements on AI-powered systems. However, a persistent gap remains between existing XAI methods and society's legal requirements, leaving practitioners without clear guidance on how to approach compliance in the EU market. To bridge this gap, we study model-agnostic XAI methods and relate their interpretability features to the requirements of the AI Act. We then propose a qualitative-to-quantitative scoring framework: qualitative expert assessments of XAI properties are aggregated into a regulation-specific compliance score. This helps practitioners identify when XAI solutions may support legal explanation requirements while highlighting technical issues that require further research and regulatory clarification.

Problem

• Explainable AI
• EU AI Act
• model-agnostic
• compliance
• interpretability

Innovation

• model-agnostic XAI
• EU AI Act
• compliance scoring
• explainability requirements
• qualitative-to-quantitative framework