CREATe is happy to present the first entry in our working paper series released in 2026: ‘Does this unit have a soul?’ AI-generated works, creativity research, and copyright policy by Joshua Yuvaraj, Senior Lecturer at the Faculty of Law, University of Auckland and a long time collaborator and friend of CREATe.
Click on image to download
Josh writes:
Generative artificial intelligence has consumed copyright discourse. Lawmakers, judges, and academics worldwide are racing to find a proper response in copyright law to technologies that enable wide-scale production of songs, images, text and more that closely mirror what has historically been the preserve of human intellectual effort. Much of the discourse has been about how to respond to the use of copyright-protected works to train AI models; yet what exactly we should do with works that have been generated by those same AI models is perhaps equally important, and even more perplexing.
This paper seeks to provide a theoretical basis on which to respond to that question in copyright law. It does so using creativity research – a body of scholarship comprising neuroscience, anthropology, psychology, genetic studies and more – which helps us understand what is creative, why creativity occurs and the various cognitive and social processes leading to creative output. This is important because copyright law is frequently justified on the basis that it promotes creativity; but we can only judge whether AI-generated works should be protected if we understand whether those works are indeed ‘creative’.
The paper shows how creativity research focuses both on the product and the process; how it is anthropocentric in nature; and how it generally coalesces around creative work being novel and useful. These qualifications do not easily cohere with copyright law, given copyright law does not place a value or usefulness requirement on qualifying works, nor does it require works to be novel in the same way that patent, trademark or designs law does. Nevertheless, the paper shows how creativity research can still inform copyright law’s understanding of creativity, chiefly due to similarities with the doctrine of originality which copyright law requires works to have before copyright will subsist.
The paper then examines the technical features of AI models at an abstract level, concluding that AI models are more than tools but less than fully creative. Consequently, it recommends repurposing the UK’s existing computer-generated works provisions – which allow copyright to subsist in works in respect of which there can be no human author, copyright being owned by the person who made the arrangements necessary for the creation of the work (see Copyright, Designs and Patents Act 1988, s 12(7))) to better account for the reality of AI-generated works. This type of provision better reflects the lopsided involvement of creators and models relative to conventional creativity (with AI models doing more of the ‘labour’). It is hoped that this analysis will be useful to the UK Government, which is presently considering amending the CDPA to address AI-related issues.
Given the fast-moving nature of the technology and relevant legal frameworks, Joshua encourages comments by email (Joshua.yuvaraj@auckland.ac.nz). He also has several other works in progress on creativity and copyright law which he looks forward to sharing in the near future.
‘Does this unit have a soul?’ AI-generated works, creativity research, and copyright policy
Joshua Yuvaraj
CREATe Working Paper 2026/1
Abstract
This article contends that determining whether copyright protection should apply to AI-generated works requires understanding whether artificial intelligence is “creative”. If it is a tool, works generated using AI deserve copyright protection. However, if it is genuinely creative and appropriates the locus of creative labour from human minds, then those works may deserve less or no protection. Examining creativity research, the article concludes that AI is more than a tool but not fully creative. It suggests that provisions governing computer-generated works may balance the need for some protection, without imposing excessive access costs on the public in respect of AI-generated works.
Full paper can be downloaded here.
See related CREATe team works:
- Martin Kretschmer, Bartolomeo Meletti, Luis H Porangaba, ‘Artificial intelligence and intellectual property: copyright and patents—a response by the CREATe Centre to the UK Intellectual Property Office’s open consultation’, Journal of Intellectual Property Law & Practice, Volume 17, Issue 3, March 2022, Pages 321–326, https://doi.org/10.1093/jiplp/jpac013
- Jim Brogden and Kris Erickson, ‘On the process of image creation with artificial intelligence: Rephotography using Midjourney AI’, CREATe Working Paper 2025/7, https://zenodo.org/records/16883055