The AI act meets general purpose AI: the good, the bad and the uncertain

Abstract

Moreira, Freitas, and Novais (2023) address the implications of the EU's AI Act concerning General Purpose Artificial Intelligence (GPAI). The paper examines the expansion of the AI Act's scope to include GPAI and analyzes the merits and potential flaws of this regulatory approach. The central question revolves around whether GPAI should be treated as a general-risk category or a high-risk one, highlighting the need for clarity in risk assessment. The authors call for a clear allocation of responsibilities among AI actors and emphasize the importance of cooperation. They also caution against exemptions that could create liability gaps. The study suggests the AI Act has merit in regulating a rapidly expanding field with broad applications but requires careful consideration to avoid stifling innovation while ensuring responsible AI governance. The research implies a need for balanced regulation that promotes AI development while addressing potential risks and harms.