Risk-Based Approach to EU AI Act: Benefits and Limits of Co-Regulation

The regulation of artificial intelligence (AI) is an important policy area in which a risk-based approach is increasingly being adopted (Ebers, 2024). This article explores the emergence of co-regulation - a hybrid form of regulation in which public and private actors coordinate their regulatory activities under increasingly risk-based approaches to AI risk management. Based on an analysis of emerging co-regulatory arrangements in European Union (EU) AI Act, the article argues that there are significant benefits and challenges to implementing such public-private arrangements, and discusses some fundamental conditions under which they can achieve the societal goals of safety, health and fundamental rights. In particular, it argues that the EU regulatory framework must respond and adapt to changing risk profiles and industry developments to ensure that public oversight of AI is risk-based, transparent and not dominated by industry power. This is particularly important to promote compliance and create an effective regulatory framework.