Assessing Prompt Inputs to Generative AI from Copyright Perspectives

The rapid development and proliferation of AI technologies heralds profound changes to the copyright regime. In this vein, conflicting views are being expressed regarding how to evaluate the creation of expressive works using generative AI from copyright law’s perspectives. One argues that generative AI is just another tool and should be recognized by copyright law from its historical perspectives, and thus it is natural to expand copyright protection therto, pointing to the bright future prospects including expansion of the expressive means and creative participants. The opposing side argues that the inherent randomness, indeterminacy and unintelligibility of AI’s operation makes it different from traditional tools for creation in terms of copyright, and that copyrighting AI outputs will lead to the extinction or degradation of human creativity. These debates are often characterized by a mixture of interpretive and policy perspectives of copyright law. This article aims to introduce a balanced outlook of the cases and debates on the copyrightability of AI outputs, with main focus on prompt inputs to the generative AI, thereby providing a basis for further discussion. Reviewed in the article are as follows: First, while the U.S. Copyright Office has been denying creative contribution by prompt inputs to generative AI, Beijing Internet Court of China and Japanese government take a flexible approach. Also, more and more scholars are turning in favor of copyrighting some AI outputs. Second, the wide spectrum of forms, contents, and use cases of prompt inputs makes it increasingly difficult to maintain a rigid attitude toward the authorship of AI outputs. Third, recognizing ‘authoring by prompt inputs’ under copyright law is ultimately a matter of copyright policy. Therefore, it is necessary to distinguish between interpretive and policy discussions. Policy makers have an urgent task to collect factual evidence on the current status and prospects of generative AI technologies in creative markets, and formulate policies based on in-depth scrutiny of the facts. It is hoped that the cases and policy/academic discussions reported in this article help lay the foundation for copyright policy debates on generative AI.