Research work · 2025

The European Union Artificial Intelligence Act: Mitigating Discrimination In Artificial Intelligence Systems

Elsevier BV

Abstract

Artificial Intelligence (AI) aims to meaningfully contribute towards developments and innovation in both the private and public sectors. However, as meaningful and well-intentioned as these interventions by AI may be, they also pose significant risks of harm in the enjoyment of fundamental human rights. This thesis takes an exploratory approach to understanding discrimination in the use of Artificial Intelligence, and how, or if, the European Union Artificial Intelligence Act (EU AI Act) offers a sufficient regulatory framework for the prevention of algorithmic discrimination. It is an interdisciplinary, legal research combining a descriptive, and a case study approach to establish a nexus between AI and discrimination; particularly in answering: what constitutes discrimination under European Union law? What constitutes discrimination in the usage of AI? Is this concept of discrimination enshrined in the EU AI Act? How can AI lead to discrimination? Is the AI Act an adequate regulatory framework against discrimination in the use of AI? Does the risk-based approach of the EU AI Act sufficiently mitigate discrimination? And is there a need for further regulation of AI to protect against algorithmic discrimination? The thesis concludes with the findings that while EU AI Act is ‘the world’s first regulatory framework’ on AI, its product-safety approach, rather than a fundamental rights-based approach renders the Act an inadequate regulatory framework for the prevention of discrimination in Artificial Intelligences systems.

DOI: 10.2139/ssrn.5087870