YE, YING (2024) Protection of AI-generated Works by Copyright:
A Human-Based Perspective. Doctoral thesis, Durham University.
Full text not available from this repository. Author-imposed embargo until 01 May 2028. |
Abstract
Artificial Intelligence (AI) is a collection of algorithms that perform tasks with a certain degree of autonomy under a computational mechanism. As a form of autonomous algorithms, when AI is applied to the creation of work, it brings challenges in deciding whether and how to protect AI-generated works through copyright law. These two copyright challenges have raised international and national discussion. However, since the copyright systems are different in various countries, there is no consensus achieved yet. The current discussions about the copyright protection of AI-generated works are mainly around three important issues, authorship, ownership and copyrightability. The literature presents that authorship is currently the most discussed issue and reveals voices for and against AI authorship. AI is not a human being, which may conflict with the fundamental requirement of human authorship in copyright law. The clarification of the authorship of AI-generated works has a direct impact on the rest two questions. Due to the lack of in-depth and systematic research on the copyright protection of AI-generated works, this research adopted doctrinal and comparative law methodology to explore whether and how to provide copyright protection to AI-generated works. This research has revealed how different understandings of the roles humans and AI play in the creation of work affect the solutions to copyright issues of AI-generated works. First, this research argued that AI is not a fully autonomous creator or a mere tool, but a participant in the creation of work. Both the “training stage” and “generation stage” should be considered when discussing the copyright issues of AI-generated works. Second, by reviewing the concept of authorship, the formula for searching for authorship, and rulings of typical non-human creation cases, this research argues that the author of AI-generated works is still humans, not AI. Furthermore, AI authorship is not accepted under the current copyright regime. Third, compared to alternative approaches, copyright law is currently the best strategy for the protection of AI-generated works. Fourth, as long as there exists human contributions in the two stages of AI creation with sufficient impact on the resulting works, AI-generated works should be considered as intellectual creations. By comparing British, European and Chinese originality standards, this research argued that AI-generated works are original works and satisfy the copyrightability requirements. Fifth, through the study of recent AI rulings, this research has proposed a more comprehensive and practical approach, the combination of the primary approach “human author approach” and the alternative approach “AI owner approach”, to give AI-generated works protection. AI-generated works can and should be protected by copyright law, with the human author approach being the most suitable framework in the era of Weak AI.
Item Type: | Thesis (Doctoral) |
---|---|
Award: | Doctor of Philosophy |
Faculty and Department: | Faculty of Social Sciences and Health > Law, Department of |
Thesis Date: | 2024 |
Copyright: | Copyright of this thesis is held by the author |
Deposited On: | 01 May 2025 12:20 |