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

TL;DR

This paper evaluates model-agnostic XAI methods against EU AI Act explainability requirements, proposing a scoring framework to assess compliance and identify gaps for practitioners.

Contribution

It introduces a novel qualitative-to-quantitative scoring framework linking XAI interpretability features to legal compliance with the EU AI Act.

Findings

A scoring framework to assess XAI compliance with EU regulations

Identification of technical gaps in current XAI methods for legal explainability

Guidance for practitioners on selecting XAI methods for regulatory compliance

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.