Guest Post by Meshandren Naidoo and Dr. Christian E. Mammen
A world first – South Africa recently made headlines by granting a patent for ‘a food container based on fractal geometry’ to a non-human inventor, namely an artificial intelligence (AI) machine called DABUS.
Over the past three years, the AI algorithm DABUS (short for Device for the Autonomous Bootstrapping of Unified Sentience) and its team of supporting humans, including Dr. Stephen Thaler and Prof. Ryan Abbott, have made headlines around the world as they sought patent protection for a fractal-inspired beverage container (shown below) that they contend was invented by DABUS.
Notably, their application has been denied by the United States Patent and Trademark Office (USPTO), the United Kingdom Intellectual Property Office (UKIPO), and the European Patent Office (EPO). The grounds for rejection have included a mix of procedural formalities, formal legal requirements, and theoretical objections. The procedural formalities and formal legal requirements, which have been equally important to the theoretical questions in these decisions, are sometimes overlooked in the popular media. They include such issues as whether (and how) Dr. Thaler obtained authorization from DABUS to file the patent application, and whether the patent statutes include a requirement that inventors be human. Each of these three jurisdictions found sufficient reasons in these formalities to reject DABUS’ patent applications. In addition, the EPO focused on the broader question of legal personhood: namely, that a number of other rights and obligations are attendant upon being an inventor, and unlike humans, an AI lacks the legal personhood to discharge those obligations and exercise those rights. The UK courts reasoned similarly, noting that an AI lacks the capacity to hold property, and therefore could not have authorized Dr. Thaler to act on its behalf. The USPTO emphasized that under US law invention requires “conception” followed by reduction to practice, and reasoned that “conception” requires a theory of mind that is simply not established to be present in an AI.
Critics of those decisions have emphasized the role of patenting as a part of national industrial policy, and in particular the role of patent grants in encouraging innovation. With increasingly capable AI algorithms, the argument provides, the ability to innovate is shifting from an exclusively human domain to one that includes the algorithms, and modern industrial policy needs to encourage and reward that shift.
These same factors appear to have come into play in the South African decision, though to a clearly different outcome.
The pitfalls of formal examination in South Africa
In July 2021, South Africa’s patent office, the Companies and Intellectual Property Commission (CIPC), granted the South African DABUS patent application, which was published in the South African Patent Journal. Unlike the USPTO, UKIPO and EPO, the CIPC does not conduct a more thorough interrogation of patent applications, known as substantive search and examination (SSE). Instead, all that is required in a formal examination (also known as a registration-based system) is for the application forms and fees to be in order with the specification documents attached. If these affairs are in order, the patent will summarily be granted by the CIPC. This, along with the lack of information provided by the CIPC post-grant has led to criticism directed towards its non-examining nature. This limited review for compliance with the procedural formalities appears to have reached a different outcome than the USPTO, UKIPO and EPO, finding that Dr. Thaler is empowered to apply on behalf of DABUS. No further information has thus far been given by the CIPC relating to the grant. It should be noted however that the ongoing patent reform in South Africa provides for training and infrastructure upgrades to accommodate a shift towards implementing SSE.
Does substantive South African patent law preclude AI inventorship?
The South African Patent Act 57 of 1978 (Patent Act) does not define an ‘inventor’ hence it is arguable that the Patent Act could, or should, be interpreted to include AI. However, the Patent Act presents some challenges in doing so such as, inter alia, the requirement for the provision of names and addresses of inventors—the EPO cited a similar requirement in denying DABUS’ application. If the reasoning of the USPTO is followed, a further challenge to the DABUS patent in South Africa would be the ‘first and true inventor test’. Like the ‘conception’ test in American patent law – the object of the test is to determine the identity of the ‘devisor’ of the invention. With that said, it is open for the South African legal system to determine if the test, which was originally crystallised in South African law in 1902 (with not much development taking place between then and now) is a bar to AI inventorship. Whilst case law which explains the test also refer to pronouns such as ‘he’ or ‘she’, South Africa could diverge from the USPTO and employ a more purposive approach to its interpretation (which is the Constitutionally recognised manner of interpretation) as opposed to a more textual one. This statutory interpretation method would include broader considerations in the process such as (1) advantages posed by AI inventorship; (2) policy directives; (3) the fact that AI inventorship was unlikely to have been considered during the period when the test was originally developed and; (4) the object of the test is to determine the identity of the deviser of the invention in the event of disputes and the like – not to preclude other non-human entities from innovating.
Was granting the patent a mistake?
At first glance, it may appear that the DABUS patent was erroneously granted by the CIPC. Although there has been a shift towards digitization, the CIPC has struggled extensively in the past with infrastructure and administrative issues. But it may be premature to conclude that the granting was erroneous. The post-apartheid government foresaw the challenges associated with the exclusion of a large portion of citizens from economic participation, and central to the solution was science, technology, and innovation. This culminated in the White Paper on Science and Technology in 1996. Soon after, came many other strategic policies aimed at placing South Africa and its citizens in a stronger position. In 2019, the Presidential Commission on the Fourth Industrial Revolution and an updated White Paper on Science and technology was published – both of which highlighted the need for a technology-orientated approach to solving socio-economic issues.
Unfortunately, innovation (noted in the 2019 White Paper on Science, Technology, and Innovation) as measured in products produced and patent output from South African applicants in the country and in other jurisdictions via the Patent Cooperation Treaty (PCT) has remained ‘relatively flat’. Adding to this was policymakers’ concerns of ‘the brain drain’ – the emigration of skilled persons in search of better opportunities and environments.
In April 2021, a call for public comments on the proposed National Data and Cloud Policy in terms of the Electronic Communications Act 36 of 2005 highlighted three main points: (1) the South African Government aims to create an AI institute to assist with reformation; (2) the intention of this is to encourage investment in, and exploration of, AI as a means to achieve sustainable development goals and economic growth; and (3) AI is viewed as a solution to some of the capacity issues facing South Africa.
Thus, as a matter of national industrial policy, it is entirely possible that the grant of DABUS’ patent is fully consistent with the emphases on broad access, digital innovation, and support of science and technology generally.
An opportunity for South Africa?
Given that South Africa is currently undergoing major patent reform, South Africa’s policymakers may find that it would be prudent to capitalise on any presented advantages. Support for, and recognition of, AI inventorship could make South Africa an attractive option for investment and innovation and may also cause these systems to be viewed as a sustainable form of innovation. The path forward for South Africa is uncertain, but there are opportunities in recognising AI as an inventor that could aid in achieving the national policy goals. In doing so, South Africa may champion the Fourth Industrial Revolution and signal leadership to other countries. Indeed, in just the few days since the South African DABUS patent was granted, the Australian Federal Court appears to have followed suit, overturning a rejection of DABUS’ application by that country’s patent office and finding that recognizing AI inventorship would be “consistent with promoting innovation.”
 Hay v African Gold Recovery Co 1902 TS 232 p 233.
 University of Southampton’s Applications  RPC 567 (CA) paras 22–25.
 Bertie Van Zyl (Pty) Ltd v Minister for Safety and Security 2010 (2) SA 181 (CC) para 21.
 Josh Taylor, “I’m sorry Dave I’m afraid I invented that: Australian court finds AI systems can be recognised under patent law,” The Guardian (July 30, 2021) (https://www.theguardian.com/technology/2021/jul/30/im-sorry-dave-im-afraid-i-invented-that-australian-court-finds-ai-systems-can-be-recognised-under-patent-law)
Mr Meshandren Naidoo is a Ph.D Fellow at the University of KwaZulu-Natal (South Africa) and member of the African Health and Research Flagship. His areas of interest are AI technology, intellectual property, business strategy, and bioethics. His Ph.D involves looking at the challenges posed by AI technology to South African patent law and what the potential solutions may be.
Dr. Christian E. Mammen is an IP litigation partner with Womble Bond Dickinson in Palo Alto, CA. He has practiced in San Francisco and Silicon Valley for over 20 years, and has held visiting faculty positions at a number of universities, including Oxford University, UC Berkeley Law School, and UC Hastings College of the Law. He has written and spoken extensively on AI and patent law, including “AI and IP: Are Creativity and Inventorship Inherently Human Activities,” 14 FIU L. Rev. 275 (2020).
The authors confirmed they did not receive compensation for this article and that they do not represent any clients that might be impacted by the article or the underlying decisions. The views expressed by them in this article are solely their own.