Introduction

Artificial intelligence (AI) is the ability of a computer to perform tasks in a way that simulates human intelligence – that is, to mimic the way humans would act, reason, learn or problem-solve. There is no universal definition of artificial intelligence. Generally, it is focused on developing machines and systems that can carry out tasks considered to require human intelligence. Two subsets of AI are “machine learning” and “deep learning” and, with the development of new neural network techniques and hardware, AI is often a synonym for “deep supervised machine learning”.

In Australia, as in many other jurisdictions, AI-generated works and inventions are unlikely to be protected by patent law and copyright law because of the absence of a human inventor or author – which is a necessary requirement in order to qualify for protection as an intellectual property right (IP).

In light of this context, the World Intellectual Property Organization (WIPO), the specialised agency of the United Nations which heads the development of global IP systems, ¹ has been leading an initiative to identify policy issues relevant to the regulation of intellectual property rights related to AI.

The issues which are being identified through this process provide insight into the treatment of AI-generated works and inventions under Australian law. Dentons Australia Associate, Tom Johnston and Partner, Robyn Chatwood, discuss WIPO’s efforts below and some of the interesting issues.

Background to WIPO’s Conversation on IP and AI

Amidst the potential to use AI to generate works and inventions without human intervention in a growing number of commercial contexts, WIPO held in September 2019 its first “Conversation on IP and AI”, bringing together Member States and other stakeholders to discuss the impact of Al on IP policy, with a view to “… collectively formulating the questions that policymakers need to ask”..²

In December 2019, WIPO released a draft issues paper for a public consultation on AI and IP policy (Issues Paper), and invited feedback about the issues to “… help define the most-pressing questions likely to face IP policy makers as AI increases in importance”. ³

Following the initial session in September 2019, WIPO hosted another session with Member States and a range of stakeholders in July 2020 to discuss the concepts set out in its Issues Paper. In particular, the sessions have sought to identify the nature of any IP rights in AI-generated works and inventions and to highlight the complications involved in determining the owners of these rights.

Member States and corporate stakeholders, including IP Australia, provided written interventions to WIPO that addressed the Issues Paper. ⁴ In May 2020 a revised issues paper was issued. ⁵

The final session in the Conversation will be conducted in November 2020.

The overlap between AI and IP

As noted above, AI involves the use a range of computer algorithms to implement techniques, such as logic programming, machine learning, deep learning and neural networks, in order to perform tasks across various applications without human intervention.

AI can be implemented, for example, in facial recognition systems that verify a person’s identity by extracting and analysing elements of facial features from a database of images. Similarly, AI is also used in self-learning smart home products that attempt to anticipate changes in temperature in order to optimise heating within a user’s home. AI has also been used to print artworks, create literary works and arrange musical compositions without human intervention.

As AI technology is increasingly adopted in order to transform business processes the question of the interaction between AI and intellectual property laws is becoming increasingly relevant, particularly as businesses attempt to restrict the use of and commercialise AI-generated works and inventions.

The current approach to AI and IP under Australian law

In the meantime, Australia’s Minister for Industry, Science and Technology released a discussion paper in April 2019 on how Australia should design, develop, deploy and operate AI, along with a set of AI ethics principles, ⁶ which were subsequently revised. ⁷

In terms of IP, Australian law does not currently recognise AI as the author of copyright subsisting in AI-generated works. Similarly, AI is not recognised as the inventor or owner of AI-generated inventions. More broadly, the ownership of intellectual property in AI-generated works and inventions is not expressly addressed in Australia’s IP legislation.

• Copyright:
  • Section 32 of the Copyright Act 1968 (Cth) provides that copyright subsists in an original literary, dramatic, musical or artistic work where the work is unpublished and the author is an Australian citizen or resident.
  • Decisions applying section 32 have developed a body of case law which requires copyright works to be created by a human author. In particular, in IceTV Pty Ltd v Nine Network Australia Pty Ltd (2009) 239 CLR 458, the High Court held that in order for copyright to subsist in a literary work, the work must have been created by a human author who has exercised sufficient mental effort in the creation of the work.
  • In Telstra Corp Ltd v Phone Directories Co Pty Ltd (2010) 201 FCR 173, the Federal Court found that copyright did not subsist in telephone directories that had been generated by an automated process using computer software, such that the persons involved did not invest any independent intellectual effort that was directed to the material form of the relevant works.
  • Historically, one of the primary reasons that Australian courts have prioritised the importance of a human author is to ensure that Australia’s copyright laws recognise the skill and labour invested by creators and to incentivise the production of artistic, literary, dramatic and musical works.

• Patents:
  • In relation to AI-generated inventions, section 15 of the Patents Act 1990 (Cth) provides that a patent for an invention may only be granted to a legal person.
  • It is also necessary to identify the inventor in order for a patent application to be granted in Australia. ⁸

WIPO Conversations and Australian interventions

The Issues Paper and the interventions provided by Australia provide an insight into the policy issues that may shape the future application of Australia’s IP laws to AI-generated works and inventions, particularly as AI is increasingly adopted in a range of commercial applications.

For context – Australia is not alone in grappling with these issues. A range of international stakeholders made submissions in response to the Issues Paper. These submissions have primarily focused on copyright law and patent law; particularly the ownership of any intellectual property rights in AI generated inventions and works.

The Issues Paper queries who should be named as the inventor in respect of AI-generated inventions given that AI lacks legal personality. WIPO also notes that if it is necessary to identify a human inventor, it remains unclear who should be named as the inventor. For example, this may be one or more of the stakeholders involved, such as the original programmer, an investor or the end user.

Separate to the identity of the inventor, WIPO also highlighted the difficulty in assessing the inventiveness of AI-generated inventions and the limitations of comparing an AI-generated invention to the approach that would have been taken by the person skilled in the art to determine if the invention is obvious.

The Issues Paper also explores the question of whether AI-generated works are original and, if so, who should be attributed as the author of the AI-generated works. In addition to the subsistence of copyright, the Issues Paper also highlights the difficulty in identifying the person responsible for the infringement of copyright if AI produces a work that infringes copyright in an existing work, particularly in machine learning contexts where an AI system may have analysed a large database of copyright materials.

In response to the Issues Paper, IP Australia commented that the attribution of copyright to AI may have a negative impact on human creative industries and dilute the incentives for human creators to produce original works. ⁹

The current state of play and the future of AI under Australian law

Patent applications have recently been filed in various jurisdictions internationally, including Australia, which name DABUS (an AI system) as an inventor in respect of a patent titled ‘Food container and devices and methods for attracting enhanced attention’ (DABUS Application).

The patent application is related to The Artificial Inventor Project which seeks to promote intellectual property rights for the autonomous output of artificial intelligence. The DABUS Application is yet to be examined in Australia. However a related application was recently refused in the United Kingdom. ¹⁰

While IP Australia’s approach to the DABUS Application may shed light on the patentability of AI-generated inventions, AI-generated inventions are, in our view, unlikely to be patentable in Australia. Ultimately there is still a significant policy discussion needed to address the application of Australia’s IP laws to AI-generated works and inventions. In the interim, businesses that have adopted AI should consider protecting AI-generated works and inventions through contractual provisions and technical measures, such as encryption, passwords and firewalls, to limit the unauthorised reproduction and use of AI-generated materials.

Next steps

We will keep you updated on the issues as they evolve. If you need advice on any aspect of AI and intellectual property rights, please contact Associate, Tom Johnston and Partner, Robyn Chatwood or your usual contact in the Dentons Intellectual Property and Technology team.

1. Established in 1967 by the WIPO Convention ↩
2. See WIPO Conversation on Intellectual Property and Artificial Intelligence ↩
3. See WIPO Begins Public Consultation Process ↩
4. See Draft Australian Government Response ↩
5. See Revised Issues Paper ↩
6. See Australia's Ethics Framework ↩
7. See AI Ethics Principles ↩
8. Patents Regulations 1991 (Cth), reg 3.2C(2)(aa) and reg 4.1 ↩
9. See Australian Government Response ↩
10. Thaler v The Comptroller-General of Patents, Designs And Trade Marks [2020] EWHC 2412 ↩