The EU AI Act is a Good Start But Falls Short

Abstract

The EU AI Act (EU AIA) was created to ensure ethical and safe Artificial Intelligence development and deployment across the EU. Enterprises will face stringent requirements when complying with the EU AIA, which many organizations currently find confusing. This study aims to identify key challenges and strategy development that organizations face regarding the EU AIA and to help them comply with it. To achieve this aim, we conducted a Multivocal Literature Review (MLR) to explore the sentiments of both the industry and the academia. From 130 articles, 56 met our criteria. We found 199 primary keywords in the literature. The three following concepts were the key ones: liability, discrimination, and a adequacy of tool. Additionally, both industry and academia expressed negative sentiments around regulatory interpretations, specific requirements, and transparency issues. Based on our findings, we highlight three themes for enterprises. First, a risk-based regulatory compliance system, emphasizing the importance of developing systems that ensure adherence to regulatory requirements by focusing on risk management. Second, the need for ethical frameworks and principles in technology development. Third, the importance of policies and systems regulatory risk management. These results highlight key challenges and strategies, while also highlighting less commonly discussed themes, thus enabling enterprises to align their systems with the requirements and better position themselves in relation to the EU market.