A critical analysis of the intersection between copyright law and artificial intelligence
Loading...
Files
Date
2024
Authors
Scannell, Barry
Journal Title
Journal ISSN
Volume Title
Publisher
University College Cork
Published Version
Abstract
This thesis examines the complex and evolving relationship between artificial intelligence (AI) and copyright law, focusing on two main aspects: AI's use of copyrighted works for training purposes (AI inputs) and AI's generation of new works that may be eligible for copyright protection (AI outputs).
The central research question of this thesis is: To what extent do existing international, regional and national copyright approaches address the challenges posed by AI-generated works, and how should these frameworks evolve to ensure harmonisation, protect human creativity, and ensure an appropriate balance between protecting human creativity and fostering innovation?
To answer this question, the thesis adopts a comparative approach, analysing legal frameworks and challenges in the European Union, the United States, and selected Asian jurisdictions. It highlights the fragmented international approaches to copyright and AI training processes (inputs) and advocates for a harmonised global framework. In addressing the complexities of authorship and ownership of AI-generated works (outputs), the thesis suggests that a possible solution may lie in the introduction of sui generis protections tailored to the unique challenges posed by AI.
This thesis is structured around three main chapters, in addition to the introduction and conclusion.
Chapter two addresses the question: Can copyright law keep up with AI training needs? It explores the phenomenon of text and data mining (TDM), a core technique through which AI systems learn from vast datasets, many of which contain copyrighted material. This chapter discusses the various intellectual property rights implicated by TDM, including the reproduction right, adaptation right, database right, and rights management information. It evaluates the adequacy of existing and proposed copyright exceptions and limitations for TDM across jurisdictions, analysing whether they balance the interests of creators with those of AI developers. This chapter also highlights the challenges posed by international disparities in TDM exceptions, underscoring the need for harmonised global approaches to AI training.
The third chapter considers the questions: What do courts say about AI and copyright? and What are the emerging legal challenges in AI infringement disputes? It provides a critical analysis of judicial decisions relevant to the emerging trend of copyright infringement lawsuits involving AI. Given the United States' advanced stage in AI-related litigation, this chapter focuses on key cases such as Authors Guild v. Google and explores the role of fair use as a central defence for AI companies. European approaches to similar copyright issues are also examined, with a focus on infringement and the broader implications for copyright frameworks. This chapter reveals the growing importance of judicial reasoning in shaping how copyright law responds to AI technologies.
Chapter three also examines the new wave of litigation brought against AI developers, exploring how key elements of such disputes such as memorisation, substantial similarity, and the use of copyrighted data are tested within current legal frameworks. This chapter evaluates how these disputes impact copyright law and identifies areas where legal clarity is needed to accommodate AI’s rapid development.
Chapter four explores the question: Who owns AI-generated works under existing copyright law? It provides a detailed analysis of copyright protection for AI-generated works, beginning with an examination of the spectrum of AI assistance, from AI as a mere tool to its role as a potential author. This chapter reviews international and national legal approaches to authorship, originality, fixation, and ownership, focusing on jurisdictions such as the European Union and the United States. Particular attention is given to US case law, where AI authorship is being actively debated, and EU jurisprudence on originality, which offers insight into how the EU may approach these issues. By addressing the challenges of defining authorship for AI-generated works, this chapter contributes to the broader discourse on adapting copyright law to account for AI’s role in creative processes.
The concluding chapter addresses two key questions: Should AI-generated works get new legal protections? and Is global harmonisation of copyright frameworks possible? It explores whether new forms of legal protection, such as sui generis rights, are needed to address the gaps in existing frameworks. The chapter also considers how such protections might be structured, including their scope, duration, and economic implications. Additionally, it evaluates the strengths and limitations of the EU AI Act in addressing copyright concerns and explores whether Ireland’s legislative approach could serve as a model for broader EU reforms. Finally, this chapter examines the major obstacles to achieving global harmonisation and argues that international collaboration is essential for fostering consistent and equitable standards for AI-generated works.
The conclusion of this thesis argues that while existing international, regional and national copyright approaches provide a foundation, they are insufficient to address the complexities introduced by AI. Harmonised TDM regulations, the introduction of transparency and accountability mechanisms, and tailored sui generis protections are proposed as solutions to balance the needs of creators, developers, and the broader public interest. By advocating for a balanced and adaptive legal framework, this thesis provides a pathway for copyright law to evolve in step with the realities of the AI era, promoting innovation while safeguarding human creativity and the integrity of copyright systems worldwide.
Description
Controlled Access
Keywords
Artificial intelligence , AI , Copyright , TDM , Text and data mining , AI Act , Generative AI , ChatGPT , Transformer
Citation
Scannell, B. 2024. A critical analysis of the intersection between copyright law and artificial intelligence. PhD Thesis, University College Cork.