Copyright in AI-generated works: Lessons from recent developments in patent law
Authors
Jyh-An Lee** Rita Matulionyte*
*Senior Lecturer, Macquarie Law School, Sydney, Australia, rita.matulionyte@mq.edu.au.
** Professor, Faculty of Law, The Chinese University of Hong Kong, Hong Kong SAR.
An earlier version of the paper was presented in the 13th IP Conference – Innovation, Intangible Assets During and After the Global Pandemic held by the Chinese University of Hong Kong Faculty of Law. The autors thank Albert Wai-Kit Chan, Peter Yu and anonymous reviewers for their helpful comments
In Thaler v The Comptroller-General of Patents, Designs and Trade Marks (DABUS), Smith J. held that an AI owner can possibly claim patent ownership over an AI-generated invention based on their ownership and control of the AI system. This AI-owner approach reveals a new option to allocate property rights over AI-generated output. While this judgment was primarily about inventorship and ownership of AI-generated invention in patent law, it has important implications for copyright law. After analysing the weaknesses of applying existing judicial approaches to copyright ownership of AI-generated works, this paper examines whether the AI-owner approach is a better option for determining copyright ownership of AI-generated works. The paper argues that while contracts can be used to work around the AI-owner approach in scenarios where users want to commercially exploit the outputs, this approach still provides more certainty and less transaction costs for relevant parties than other approaches proposed so far.
