Navigating Regulatory Landscapes: A Qualitative Review of Compliance Challenges in High-Risk AI Applications Post-EU AI Act

Abstract

The increasing deployment of artificial intelligence (AI) in sensitive domains has prompted regulatory bodies within the European Union to introduce the EU AI Act, aiming to establish robust safeguards for high-risk applications. While the Act marks a significant step in AI governance, its implementation has exposed persistent compliance challenges across legal, technical, and institutional contexts. This study investigates these challenges through a qualitative review, focusing on how diverse stakeholders—regulators, developers, and end-users—navigate the evolving regulatory landscape. Using a qualitative literature review method, this research integrates findings from peer-reviewed articles, policy papers, and regulatory analyses indexed in international academic databases. Data collection was conducted through purposive sampling and thematic extraction, while data analysis followed a three-phase coding process comprising open coding, axial coding, and thematic synthesis. The study identifies fragmented regulatory interpretations among EU member states, a lack of institutional preparedness for enforcement, and the absence of agile audit mechanisms capable of addressing dynamic AI models. Furthermore, small and medium-sized enterprises face disproportionate barriers to compliance, including high administrative burdens and limited access to regulatory expertise. Variability in risk classification and definitional ambiguities complicate consistent adherence to the EU AI Act. The findings highlight the urgent need for harmonised compliance guidance, sector-specific enforcement strategies, and increased institutional capacity. Enhanced stakeholder collaboration and integrating adaptive regulatory tools are essential to ensure innovation and accountability. Future research should explore experimental governance models and cross-sector compliance best practices to inform ongoing policy refinement.