Should AI Generated Work be Subject to Copyright Law?
In recent times, Artificial Intelligence, or ‘AI’, has entered the public consciousness and is being employed for mainstream use in a significant way. Tools like ChatGPT and Stable Diffusion can be used to write an academic essay, create artworks, generate pieces of creative writing, or even answer personal questions like constructing a daily routine or travel itinerary.
AI is a branch of computer science focused on building systems that can process vast amounts of complex data and synthesise it to make decisions only humans usually would. AI systems use techniques such as machine learning, deep learning, and natural language processing to analyse and interpret data, and make predictions or decisions based on that data.
This has created unique challenges to the Australian Intellectual Property (“IP”) Regime as, under section 31(1) of the Copyright Act 1968 (Cth) (“the Copyright Act”), protections only subsist where the works are an original expression from an author who is a qualified person (Australian Citizen). Over time, the common law has established that only works produced by a human author may be original.
The issue of authorship for AI-generated works can be complex, as it is unclear who should be considered the creator of the work. In some cases, the AI system may be considered the author, while in others, the individual who programmed the AI or who owns the system may be considered the author. This issue can have significant implications for the ownership and exploitation of AI-generated works.
Moreover, the use of AI in the creation of copyright works raises questions about the responsibility of the parties involved. For example, in cases where AI systems are used to generate works based on existing copyright works, questions arise about the extent of the user’s liability for copyright infringement. Similarly, the responsibility of AI system developers and owners for the actions of the systems they create, or control may also be unclear.
The issues surrounding authorship, copyright law and AI generated works are, as such, ambiguous and difficult to detect. For example, did you realise the previous two paragraphs were written by ChatGPT?
Thus far the law has been unable to answer these complex questions. The leading authority on these issues is the case IceTV Pty Ltd v Nine Network Australia Pty Ltd, which states the work must be created by a human author who has exercised independent intellectual effort. This was endorsed in case Telstra Corp Ltd v Phone Directories Co Pty Ltd, as the Federal Court held that the white pages and yellow pages phone directories were not copyright works, since they were compiled by an automated process using computer software. However, technology has progressed substantially since 2014 as computers have transgressed the boundary of just a tool used to create works, to being able to learn and produce works without human input.
In Acohs Pty Ltd v Ucorp Pty Ltd, the Federal Court held that copyright did not subsist in source code where it was generated by a computer program. As a result, a work created by AI ends up devoid of copyright in Australia because there is no identifiable human author.
The Australian IP regime is designed to facilitate innovation and promote the commercialisation and the sharing of new works. A significant number of resources are invested into creating an AI algorithm and the outputs generated by AI systems can be a valuable asset. Hence, failing to provide copyright protection for the output of AI systems will increase risk and lessen investment, and in turn, the uptake of AI in Australia. This effectively stifles innovation, and in turn, directly contradicts the principal motive of copyright law.
Therefore, it is important for the law to amend this failure to facilitate adequate protections for these ever-growing AI generated works. I suggest two different avenues that may effectively achieve this:
- Firstly, an amendment to section 32 of the Copyright Act which specifies the author of a computer-generated work as “the person by whom the arrangements necessary for the creation of the work were undertaken.” This is based on similar provisions recently enacted in the UK and New Zealand. The UK case Nova Productions Ltd v Mazooma Games Ltd exemplifies the success of this, as a primary programmer obtained the copyrights of the AI generated work.
- Alternatively, the AI system itself could be determined as the author and owner of the copyright. This would require an amendment to section 32 of the Copyright Act to include a non-human entity as an ‘author.’ This has been suggested internationally, for example, Saudi Arabia grants special status in the form of citizenship, and the European Parliament has proposed a status called ‘electronic person.’
Immediate reform to the Copyright Act extending its protections to AI-Generated works is essential as these technologies are only inclined to become more accessible and incorporated in society. Legislative change is vital to keep up with this evolving environment.
 Stephan De Spiegeleire, et al, ‘What is Orifical Intelligence?’ (2017) Hague Centre for Strategic Studies 25, 28.
 Copyright Act 1968 (Cth) s 32(1).
 IceTV Pty Ltd v Nine Network Australia Pty Ltd (2009) 254 ALR 386, 33, 151
 Telstra Corp Ltd v Phone Directories Co Pty Ltd (2014) 316 ALR 590, 680-681.
 Shubhi Trivedi, ‘Implications of Artificial Intelligence on Copyright and Patent’ (2019) 1(4) LexForti Legal Journal 48, 49.
 Acohs Pty Ltd v Ucorp Pty Ltd (2012) 287 ALR 403.
 Andrew Wiseman and Bryanna Workman, ‘Copyright in the age of artificial intelligence and authorless works’, Australian Intellectual Property Law Bulletin (LexisNexis Australia, 2019) 34, 36.
 Ibid 34.
 Nova Productions Ltd v Mazooma Games Ltd (2006) All ER (D) 131, 105.
 Shubhi Trivedi, ‘Implications of Artificial Intelligence on Copyright and Patent’ (2019) 1(4) LexForti Legal Journal 48, 50.