| Artificial intelligence (AI) or "machine learning" poses a serious challenge to the boundary of fair use due to its massive use of data, including copyrighted works. In order to mitigate the concerns about massive copyright infringement and promote the development of data mining and AI learning, many countries have made legislative responses to this challenge. Although China has adopted the "three-step test" in its newly-revised Copyright Law, this test provides only some "semi-closed" limitations and exceptions, rather than the necessary legal mechanism for accommodating the massive "learning" and use of copyrighted works by AI. Reflection from the holistic perspective of structuralism shows that, the traditional copyright law, consisting of broad reproduction rights and narrow "limitations and exceptions", cannot meet the need for massive use of copyrighted works in the AI era. In order to promote AI learning and technological innovation, the current "limitations and exceptions" system should be expanded and reshaped. Firstly, the term "fair use" should be adopted to replace "limitations and exceptions". Secondly, with the rise of AI creation and the gradual fading of the role of human authors, the relationship between copyright protection and fair use should be reconstructed, and the status of the fair use system should be improved. Thirdly, the massive use of copyrighted works for AI learning and creation should be covered in the scope of fair use. Only the "output" work generated by AI will fall under the control of copyright holders when it is substantially similar to the original work. Finally, it is advisable for China to adopt an elastic and flexible fair use legislative model in its copyright law. |
京公网安备11010102006664号 京ICP备2023015853号