Mapping the Regulatory Learning Space for the EU AI Act
Dave Lewis, Marta Lasek-Markey, Delaram Golpayegani, Harshvardhan J. Pandit
TL;DR
The paper analyzes the EU AI Act as a landmark, horizontally-applied AI regulation that creates substantial regulatory uncertainty around fundamental rights and cross-sector enforcement. It proposes a parametrised regulatory learning space to structure multiple arenas for learning among value-chain actors, oversight authorities, and affected stakeholders, and argues that learning-driven governance, supported by open standards and interoperable information exchanges, is essential to keep pace with AI advances. Key contributions include a three-axis learning-space model, a taxonomy of learning activities, and a standards-based interoperability framework (including semantic ontologies and data spaces) to coordinate regulation, compliance, and rights protections across Annex I and Annex III, while aligning with existing EU infrastructures. The practical impact lies in enabling timely, legitimate, and efficient enforcement of the AI Act, reducing regulatory burden, and informing future updates and related EU digital regulations through a shared knowledge base and coordinated learning.
Abstract
The EU AI Act represents the world's first transnational AI regulation with concrete enforcement measures. It builds on existing EU mechanisms for regulating health and safety of products but extends them to protect fundamental rights and to address AI as a horizontal technology across multiple application sectors. We argue that this will lead to multiple uncertainties in the enforcement of the AI Act, which coupled with the fast-changing nature of AI technology, will require a strong emphasis on comprehensive and rapid regulatory learning for the Act. We define a parametrised regulatory learning space based on the provisions of the Act and describe a layered system of different learning arenas where the population of oversight authorities, value chain participants, and affected stakeholders may interact to apply and learn from technical, organisational and legal implementation measures. We conclude by exploring how existing open data policies and practices in the EU can be adapted to support rapid and effective regulatory learning.