Artificial Creations, Real Rights: Examining Copyrightability of Generative AI Data

Abstract

Since the launch of ChatGPT by OpenAI, the investment in generative AI has quintupled. While individuals employing these tools are creating novel work every day, the parent companies are generating billions of revenues. With increasing number of high-profile lawsuits relating to Generative AI and Copyright around the world, the paper discusses the concept of copyrightability in the new era of technological advancements. Generative AI undergoes massive training using data to understand images, sounds and visuals. They then respond when prompted by a user. Through this paper the authors discussed the basic concepts of ownership and then examine different scenarios to determine authorship over works generated by AI. Building on established concept like the essential need of human authorship and creativity element to grant copyright to the works of AI, the paper explores and divulge into new doctrines like the “Significant Input Test”. The authors argue that while copying large amount of copyrighted content off the internet and generating new content makes out a case for copyright infringement, the generative AI algorithm is premised on “transformative use” which in turn constitutes fair use. The authors did a comparative study of judicia-legal development around the concept of copyrightability of the content produced by Generative AI focusing on jurisdictions like US and EU, suggesting a way forward for Indian legislature.