In the near future, Artificial Intelligence (AI) is perceived to become an essential part of our daily lives.1 AI technologies present opportunities for development in innovation. This is evident with “traditional human labour and investment production services” that are inevitably being replaced by machine learning applications, that have the potential of generating “welfare-enhancing innovations.”2 The progression of AI is one that prompts questions of regulation, as issues of social and economic impact come into play.3 Looking towards Intellectual Property (IP) law, the regime is argued to have “a symbiotic relationship” with the regulation of new technologies. Objectives of the IP regime are centred around encouragement of new technologies and creations through the granting of private rights over the technologies.4 The granting of rights is held to promote a “sustainable economic basis for invention and creation.”5 Appreciating this, CIPIT will publish a series of blogs aimed at exploring the applicability of IP rights within AI innovations. The first part of the series specifically looks at patent law and AI. It assesses whether AI systems can be patent inventors and owners.

AI and Patents

The protection of innovation is one that falls within the scope of patent law.6 Looking towards patent law, “in order for there to be a patent, there first must be an inventor.”7 AI has showcased the capability of innovating “patentable ideas.”8 Key illustrations of this are found in inventions such as robotic scientists. An example of this is Eve , she is an “AI endowed robot” that is designed to assist in ‘speeding up the drug discovery process through automating the established process of hypothesis led research’.9 Eve’s contributions have led to advanced discoveries that assist with fighting diseases such as Malaria.10 When perceiving the achievements of Eve, questions around the scope of patent law arise, where ambiguity is found in the applicability of patent law in AI. Looking towards the “letter of law”, there is ambiguity specifically in establishing whether AI systems can be held to be “inventors” of a patent.11 Therefore, appreciating that AI systems are capable of creating inventions that “meet statutory requirements for patentable matter”, the underlying dilemma present is whether AI systems are included within the scope of an inventor, and as a result can they be granted patent rights ?12

The above question is one that has received a fragmented response across various jurisdictions. Specifically considering the DABUS case, which we have previously discussed here.13 DABUS is held to be at the centre of an AI project, where the objective was to qualify for IP rights for inventions generated by an AI without the presence of a “human inventor”.14 To actualise the goal, Dr Thaler, the owner of DABUS aimed to pursue national patent grants for innovations such as a fractal container (food container) and a neural flame (flash light) where DABUS would be named as the inventor.15 The DABUS patent application was filed under the Patent Co-operation Treaty (PCT) international patent system seeking protection in the United States, Europe, Australia, and South Africa. Bearing this, we look at the various fragmented approaches to the patent grants.

A Fragmented Approach ?

Starting off with the European Union Patent Office (EPO), the application failed to meet the threshold of a “human inventor” as there was a refusal to include DABUS within the scope.16 The wordings of statute were solely in reference to entities with legal capacity, thus extending the scope to include DABUS would be one that goes “beyond the wording of the applicable rule.”17 Therefore, within the EPO approach , human inventorship is held to be a prerequisite in establishing the inventive steps necessary for patent rights. Hence, the stance is, it is “only products of human inventorship that merit European patents.”18

The Australian approach is one that is disintegrated. Initially the Federal Court held DABUS to be an inventor. Under section 15 of Australia’s Patents Act 1990(Ch) and Reg 3.2C of the Patents Regulations 1991(Ch), an “inventor” has not been defined.19 The court held that in the twenty-first century, we have to appreciate that “scientific discoveries” inspire “new technologies.”20Therefore, the expanding of inventorship to include non-human agencies was deemed necessary for the development of patent law.21 Furthermore, the “notion of computer inventorship “ was one that was aligned to object clause 2A of the Patents Act 1990 (Ch) as the clause stated that the patent system should aim to “promote economic well-being through technological innovations and the transfer and dissemination of technology”.22 However, upon appeal, the Australian position changed to one that denied inventorship, as the court concluded that AI systems could not be named as inventors.23 Despite ‘inventor’ not being defined, the court argued that interpretation should adopt the ordinary English meaning.24 Moreover, “the law relating to the entitlement of a person to [the] grant of a patent” is premised upon “an invention arising from the mind of a natural person or persons.” 25 Therefore, the judgement, having consideration for “ statutory language”, history and objectives of the Patents Act, held that it was not viable for AI to be granted inventor status.26

The South African approach was a departure from the ‘mainstream approach’, as the South African Patent Office (SAPO) granted a patent for DABUS.27 Considering that SAPO does not conduct substantive examination of patent applications,28 the hurdle for the DABUS patent application was the satisfaction of ‘formal requirements’ where there was only need to identify the inventor in the patent application. Looking at the legislation present, both the South African Patents Act and Regulations failed to expound on who and what constituted an “inventor.” Therefore a “purposive approach” was present where the Act and Regulations were interpreted in line with the main objectives of IP as a means of “promoting innovation, technology… and more broadly, economic growth.” 29 Hence, “given the reality of AI technology” and its achievement of having systems “capable of autonomous invention”, the decision by SAPO was held to be progressive in nature.30

Way Forward ?

Observing the various approaches, the claim for inventorship and ownership are limited by interpretation of statute. Assessing the justifications of IP rights, the approach taken is one that includes the ‘nexus factor’ of human intervention, as IP rights are held to be ‘exclusively reserved for humans’.31 This is evident with various traditional theories such as the labour theory, that focuses on quantity of labour produced as well as the personality theory that touches upon the quality of labour, and lastly the reward theory that is based on the “utility of the output”.32 Appreciating IP’s core function as one that promotes innovation, the preservation of traditional justifications is ‘retrogressive’.33 The “romantic image” of “human ingenuity” is held to be made redundant by market realities and the needs of the modern day.34 Thus, questions of whether there is need for remodification or the replacement of the IP regime come into play, as focus here is shifted from “inventors” to “investors” of the market.35

Illustration by Michele Marconi

1 Bosher H, Gurgula O, Stokes S, Wang F, Weslenberger P, WIPO Impact of Artificial Intelligence on IP Policy Response [2020] Brunel University London, Law School & Centre for Artificial Intelligence, 4.

2 Hitly R, Hoffmann J, Scheuerer S, Intellectual Property Justification for Artificial Intelligence <> last accessed 26th June 2022.

3 Bosher (n1).

4 WIPO Magazine , Artificial Intelligence and Intellectual Property : An interview with Francis Garry <> last accessed 28th June.

5 ibid.

6 Afshar S M, Artificial Intelligence and Inventorship – Does the Patent Inventor Have to be Human? [2022] HSTLJ Vol 13:5, 56.

7 Kovach B, Ostrich with its Head in the Sand : The Law, Inventorship and Artificial Intelligence [2021] JNJTIP Vol 19:1, 143.

8 ibid.

9 Lenton D, AI scientist Eve finds toothpaste ingredient may help fight malaria<> last accessed 22nd July 2022.

10 ibid.

11 ibid.

12 Afshar(n7)59.

13 The DABUS Patent: Kenya what say ye? Can we or can we not patent AI?


14 Oriakhogba O D, DABUS gains Terriotry in South Africa and Australia: Revisiting The AI-Inventorship Question [2021] SAIPLJ 9,3.

15 ibid..3-4.

16 EPO Grounds for Refusal: In Re EP 18 275 163.6, 27 Jan. 2020 (fractal container) <>3 last accessed on 28th June 2022

17 ibid.

18 Stankova E,Human inventorship in European patent law [2021] CLJ 80(2), 338.

19 AUIPO Decision of 9 February 2021 – Stephen L. Thaler [2021] APO 5 ; Thaler v Commissioner of Patents [2021] FCA 879.

20 ibid.

21 ibid.

22 ibid.

23 Commissioner of Patents v Thaler [2022] FCAFC 62.

24 ibid.

25 ibid.

26 ibid.

27 The Conversation, In a world first, South Africa grants patent to an artificial intelligence system,

<> last accessed 29th June 2022.

28 Berger J & Rens A, Innovation and Intellectual Property in South Africa : The Case for Reform,

<> last accessed 22nd July 2022 ; “ South Africa does not conduct substantive examinations before granting patent protection. This is despite the fact that section 34 of the Patents Act appears to contemplate the establishment of a substantive patent examination system.”

29 South African Department of Trade and Industry (DTI). Intellectual Property Policy of The Republic of South Africa Phase 1 2018. Pretoria: DTI; 2018. 

30 Thaldan D &Naidoo M, AI Inventorship : The right decision? [2021] S.Afr.J. Sci , vol 117: 11/12, 2.

31 Jhy-An L, Reto H, Kung-Chung L, Artificial Intelligence and Intellectual Property ; Reto H M, Hoffmann J and Scheuerer S; Intellectual Property Justification for Artificial Intelligence (Oxford Scholarship Online 2021).

32 ibid.

33 ibid.

34 ibid.

35 ibid.

Leave a Comment

Your email address will not be published. Required fields are marked