Doctrinal Gaps in the EU Artificial Intelligence Act that Negate Effective Legal and Regulatory Oversight of Generative AI
This paper undertakes a critical examination of the European Union Artificial Intelligence Act (EU AI Act) with a specific focus on its doctrinal gaps that impede effective legal and regulatory oversight of generative AI. In an era characterized by the rapid evolution of AI, the imperative for a robust legal framework is paramount, one that not only ensures compliance with existing intellectual property laws but also anticipates the dynamic nature of generative AI. Unlike traditional AI, generative AI models analyze and curate existing information to produce novel content, a characteristic that introduces significant complexities for comprehensive regulation. While the AI Act introduces crucial provisions on transparency and risk management for general-purpose AI models, it notably refrains from explicitly defining ‘Generative AI.’ This omission, coupled with other significant shortcomings, creates lacunae that hinders its effective control over generative AI. Specifically, the Act fails to adequately address the determination of ownership for wholly AI-generated outputs, imposes limitations on disclosures that negatively impact oversight, and exhibits inconsistencies in regulating open-source models. Furthermore, it does not establish direct remedies for creators, thereby creating practical enforcement barriers. This paper systematically explores these limitations, including issues related to open-source exemptions, the absence of clear copyright ownership rules, and challenges in ensuring transparency and provenance. It also critically analyzes the Act’s enforcement mechanisms, highlighting the lack of mandated forensic capacity or centralized clearing mechanisms necessary for large-scale, practical enforcement. The discussion concludes by proposing concrete remedial measures and potential avenues for reform to strengthen the EU AI Act’s regulatory framework for generative AI, ensuring a more comprehensive and adaptable governance structure for this rapidly advancing technology.