Who holds the copyright for a work created by an AI algorithm: the person who programmed the AI or the AI itself

Abstract 

The rapid evolution of generative Artificial Intelligence (AI) challenges the fundamental,  anthropocentric basis of global copyright law, provoking a critical debate over who owns the  copyright for autonomously created works, the programmer, the user, or the AI itself.  Jurisdictions similar to United States and the European Union reserve copyright exclusively  for human intellectual creation, a stance cemented by judgments such as Thaler v US Copyright  Office, which ruled that AI systems lack the necessary human authorship. On other hand, the  United Kingdom and South Africa utilise a legal fiction, deeming the author of computer generated works to be the person who undertook the necessary arrangements for creation.  Drawing on South African precedents including the Haupt and Payen cases, this article argues  that the inherent ambiguities in defining human control versus autonomous output render  existing frameworks inadequate for highly sophisticated AI. Urgent legislative reform,  potentially through clarified human creative control criteria or the establishment of sui generis rights, is essential to balance technological advancement with the core protective goals of  intellectual property. 

Keywords: AI-generated works, copyright law, authorship, ownership, human creativity, and sui generis rights.

Introduction 

The proliferation of Artificial Intelligence (AI), particularly generative AI systems such as  ChatGPT and DALL-E, has profoundly disrupted creative sectors by generating sophisticated  content including art, literature, music, and software code.1 This technological advance  challenges long-held assumptions regarding authorship, creativity, and ownership.  Traditionally, copyright law was designed exclusively to protect the intellectual property and  labour of human creators, granting them exclusive rights over their original expression.2 

The critical issue arising today is foundational, who holds the exclusive rights to works  produced by an entity that lacks the human qualities of intent, consciousness, and legal  personhood? This dilemma holds massive economic and legal implications for all industries  dependent on protecting creative works. Historically, discussions around computer-related  works distinguished between computer-aided works, where human direction is paramount, and  computer-generated works, where machine autonomy is high.3 Modern AI often blurs this line,  producing unpredictable outputs.4 

The prevailing legal debate stems from the anthropocentric view of authorship, which generally  limits copyright to natural persons.5 This article aims to analyse the competing claims of  ownership the programmer, the user, and the AI itself under current international and  comparative legal standards, including specific judicial interpretations from South Africa and  the US. It is argued that while the human authorship requirement remains dominant, existing  legal structures are insufficient to address purely autonomous AI outputs, necessitating  prospective legislative modifications, such as the introduction of sui generis protection, to  account for technological investment. 

2. Authorship. 

2.1. Existing laws  

2.1.1The United States and the European Union 

In the United States, copyright protection is reserved exclusively for works created by a human  being.6 The US Copyright Act protects original works of authorship, and cases affirms that  copyright protects the fruits of intellectual labor founded in the creative powers of the mind.  Non-human expression is ineligible, regardless of its sophistication.7 The law is structured to  incentivise human individuals not non-human actors. 

Similar to the European Union (EU) legal framework maintains an anthropocentric focus. EU  copyright law links originality to the author’s own intellectual creation, meaning that creativity  and moral rights are attributed only to natural persons.8 As AI systems lack legal personality,  they cannot meet the requirements for authorship.9 

2.1.2 The UK and South Africa (computer generated work) 

The United Kingdom (UK) and South Africa diverge slightly by utilising a statutory fiction  intended to assign ownership for works lacking a human author.10 Section 9(3) of the UK  (CDPA) defines a computer-generated work as one created by a computer where there is no  human author.11 The author shall be taken to be the person by whom the arrangements  necessary for the creation of the work are undertaken.12 This provision acts as a legal  mechanism to support investment in the underlying system. 

South Africa’s Copyright Act 98 of 1978, as amended by the Copyright Amendment Act 125  of 1992, contains a similar provision. For a literary, dramatic, musical, or artistic work, or  computer program that is computer-generated, the author is deemed to be the person by whom  the arrangements necessary for the creation of the work were undertaken.13 For a computer  program itself, the author is defined as the person who exercised control over the making of  the computer program.14 This framework suggests that authorship could vest in the developer  or the commissioner of the AI system. 

2.1.3 International (TRIPS) 

The Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS),  administered by the World Trade Organisation (WTO), establishes minimum IP standards for  member countries. Computer programs, article 10(1) of the TRIPS Agreement mandates that  computer programs must be protected as literary works under Berne.15 This means that while  member states like South Africa, may categorise computer programs sui generis, the level and  nature of protection must align with that afforded to literary works under the Berne  Convention.16 

In authorship, the Berne Convention implicitly relies on human authorship, linking the term  of protection to the life of the author (article 7) and defining jurisdiction based on the author’s  nationality or residence (article 3). The Berne Convention also addresses moral rights, although  TRIPS article 9(1)states that members are not obligated regarding the rights conferred under  Berne Article 6 bis (moral rights). These provisions reinforce the international preference for  human creators. 

3. Judicial interpretation analysis 

3.1 United States: The Thaler17 precedent 

The definitive US judgment confirming the rejection of purely autonomous AI creations in the  case of Thaler v Perlmutter. Stephen Thaler sought to register an artwork created solely by his  AI system, DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), naming  the AI as the author or inventor.18 The US Copyright Office (USCO) denied the registration and maintaining that copyright requires human authorship, a standard affirmed by the court which deemed human creativity the sine qua non of copyrightability.19 

In the context of hybrid works, the USCO clarified in the case of Zarya of the Dawn, that  copyright protection extends only to the elements contributed by the human author (such as the  text, arrangement, and curation of visuals) and not to the AI generated images themselves.20 The USCO stressed that for a work to be protected, the traditional elements of authorship must  be determined and executed by the human, not the technology.21 

3.2 South African judicial interpretation: Payen and Haupt 

South African courts have previously dealt with the distinction between works created with  human input and those generated by computers. The pre-generative AI case of Payen  Components SA Ltd v Bovic Gaskets CC and Others,22 involved computer-produced works,  raising the question of whether the computer, or the programmer, could be the author.23 The  court concluded that the computer was no more than the tool used to produce works under  human instruction via computer programmes.24 The distinction between computer-generated  (where the real work has been done by the computer and human contribution is too trivial) and computer-aided works was recognised as valid. The idea that the programmer or operator is the  author was rejected if the computer performed the real work, as this would be factually wrong.25 

The subsequent leading South African case on software copyright, Haupt t/a Softcopy v  Brewers Marketing Intelligence (Pty) Ltd and Others,26 confirmed that a computer program is  protected by copyright in South Africa and that the author is the person who exercised control  over the making of the computer program.27 In Haupt, the court examined the working  relationship of the parties and concluded that the commissioner (Brewer Marketing  Intelligence, or BMI) exercised the requisite control over the manner in which the program was  made, thereby vesting authorship (and first ownership) in the commissioner, rather than the  developer or employee.28 This interpretation of control focuses on the power of regulation over  the maker’s work, meaning actual creative involvement is not critical.29 This precedent  reinforces that in South Africa, copyright in computer works often vests based on  organisational control or investment (the commissioner or employer). 

3.1.1 Critical evaluation 

The South African approach, especially when interpreted through Haupt’s emphasis on  managerial control, facilitates investment protection, aligning with the necessary arrangements framework.30 However, this framework applied to modern generative AI becomes highly  ambiguous.31 If the AI is truly autonomous, the link between the programmer’s initial effort or  the user’s prompt and the final output is highly attenuated making it difficult to definitively  assign control or necessary arrangements without clear legislative guidance.32 As the Payen  case suggested, if the machine does the original work, it seems factually wrong to claim the programmer is the author, now someone must hold the rights to protect the economic  investment.33 

4. Ownership and challenges 

4.1 comparative evaluation 

The programmer’s claim to ownership rests on the significant intellectual effort invested in  coding and training the original AI system. However, this connection is often viewed as too  remote, particularly when the AI’s output is unpredictable to the developer. Critics suggest that  the programmer merely provides the electronic equivalents of paints on a palette. 

On other hand, the user’s claim is strong only when they exercise sufficient independent  intellectual effort or creative control through sophisticated prompting, selection, or  modification (creating an AI-assisted work). The challenge lies in distinguishing whether a  user’s prompt constitutes genuine ultimate creative control (making the human the author) or  merely functions as instructions given to a commissioned machine (where the AI determines  the expressive elements, rendering the output non-human authored). The US Copyright Office  has maintained that providing text prompts alone is often insufficient for copyright protection. 

4.2 UK and South Africa 

The legal fiction used in the UK and South Africa, attributing authorship to the person  undertaking the arrangements necessary for the creation of the work, aims to assign ownership  where no human author exists.34 However, identifying this person for complex deep learning  models is difficult, as it could be the original programmer, the corporate owner who funded the  system, or the end user.35 Furthermore, some scholars caution that relying on this provision  risks setting a precedent for the expansion of corporate ownership of creative works, thereby  diminishing the traditional focus on the individual human creator. 

4.3 Originality 

A major legal hurdle is whether AI-generated works meet the threshold of originality.36 AI  systems generate content by analysing massive datasets, which often include copyrighted  materials.37 Critics argue that AI outputs may be merely a statistical recombination of existing  material rather than novel creations.38 If the output is simply a mechanical intellectual  achievement or if different people would get the same result, it lacks originality.39 This raises  significant concerns about unintentional plagiarism and copyright infringement, threatening to  consign unprotected AI output to the public domain. 

5. Developments 

Policy guidance in the United States affirms that generative AI outputs are copyrightable only  if a human has contributed creative, expressive content.40 Simple prompts are generally  insufficient to satisfy the standard of human authorship. In the European Union, the proposed  AI Act,41 while not directly addressing copyright ownership, focuses on transparency and  accountability.42 It mandates that providers of general-purpose AI models must implement  policies to respect EU copyright law, including rights reservations, and must make publicly  available a sufficiently detailed summary about the copyrighted content used for training.43 This highlights a growing focus on the input side of the AI creation process. In South Africa,  lacking specific AI legislation, relies on existing legal principles but policy consensus indicates  that the 1978 Copyright Act’s authors (or subsequent amendments) did not anticipate modern  AI, necessitating future legislative intervention.44 

There are no sources in the current document. 

Recent, high-profile lawsuits illustrate that the current battleground revolves around copyright  infringement based on the unauthorized use of copyrighted material for training AI models.45 Cases such as Getty Images (US), Inc v Stability AI, Inc and Silverman et al. v OpenAI, Inc.  allege unauthorised scraping of data constitutes infringement.46 These cases require courts to  assess whether fair use doctrine applies to machine learning processes, an unpredictable  determination.47 

6. Suggestions  

The creation of a new and unique set of rights (sui generis rights), separate from traditional  copyright, is a potential comprehensive solution for purely autonomous AI-generated works.48 These rights could offer protection for investment, potentially with a shorter term than human  life (e.g the UK’s 50 year term for computer-generated works),49 while avoiding reliance on  problematic concepts like moral rights for machines.50 

Legislatures, including those in the UK and South Africa relying on the arrangements necessary clause, must explicitly define the necessary level of human input (such as creative prompting,  curation, or selection) required for AI-assisted works to qualify for traditional copyright  protection.51 For autonomous works, clear rules should vest ownership in the legal entity that  owns the AI system based on an investment or proximity approach, ensuring rights are held by  a responsible party.52 

To protect human creators and mitigate plagiarism risks, legislation should mandate  transparency regarding the copyrighted training data used by AI models.53 This could be  coupled with implementing compensation models, such as revenue sharing or royalties to  ensure content creators are justly remunerated for the use of their copyrighted works in training  datasets. The judiciary must continue to delineate the boundary between mechanical assistance  and sufficient human creativity on a case-by-case basis (as seen in the partial protection granted  in Zarya of the Dawn). The legislature must introduce progressive amendments to eliminate  the inherent anthropocentric limitations where they are no longer functional. 

     7. Conclusion 

The question of who holds the copyright for a work created by an AI algorithm rests on the  evolving legal distinction between human creativity and machine autonomy. In systems relying  strictly on human authorship (US or EU), autonomously generated AI content is currently  unprotected. In jurisdictions employing legal fictions (UK or South Africa), ownership may be  assigned to the developer, owner or user based on the level of control or necessary  arrangements, as highlighted by the historical interpretation in South African cases like Haupt and the analytical distinction noted in Payen. 

The core dilemma is that traditional IP laws incentivise the personal, creative effort of a human  author, a concept that does not align with algorithmic creation. However, the economic value  stemming from the development and use of sophisticated AI systems requires protection. 

To balance innovation with protection, the future legal framework must move beyond archaic  interpretations of authorship towards pragmatic solutions. This means either refining the  definition of human authorship to encompass modern creative control (e.g detailed prompting  and curation) or adopting tailored sui generis rights for truly autonomous creations. Clarity of  ownership is essential for maintaining a fair and effective intellectual property system that  fosters innovation while protecting the rights of human creators in the rapidly accelerating  digital era.

Reference(S):

Legislations 

• Copyright, Designs and Patents Act, 1988. 
• Copyright Act 98 of 1978. 
• Copyright Amendment Act 125, 1992. 

Books 

• BARRY B SOOKMAN COMPUTER, INTERNET, AND ELECTRONIC  COMMERCE LAW (e-book) 17 (1st ed,1989). https://intl.westlaw.com/Document/Iaebc74c3421711ecb979c967a094f0c2/View/Full Text.html?ppcid=4eb1de47444a42899ff2379480a3817c&originationContext=docume nttoc&transitionType=CategoryPageItem&contextData=%28sc.Default%29. (E-book) 

Case Laws 

• Payen Components SA Ltd v Bovic CC and Others (1995) (4) SA 441 (A) (South  Africa). 
• Haupt v Brewers Marketing Intelligence (Pty) Ltd (2006) SCA 39 (South Africa) • Thaler v Perlmutter No 22-1564 (2023) (United States). 

Journal Articles 

• Anshul Kumaria & MIT World Peace University, The copyright issue in AI-generated  content: legal challenges and future directions, 6 IJL & LR 1765 (2024). 
• Enrico Bonadio & Luke McDonagh, Artificial intelligence as producer and consumer  of copyright works: evaluating the consequences of algorithmic creativity, 2 CRO 1, 1  (2020). 
• ENRICO BONADIO & LUKE McDONAGH, Artificial intelligence as producer and  consumer of copyright works: evaluating the consequences of algorithmic creativity, 2  CRO 1, 33-35 (2020). 
• João Pedro Quintais, Generative AI, copyright and the AI Act, 56 CL & SR 1 (2025).
• Julia Dickenson & Alex Morgan, et al, Creative machines: Ownership of copyright in  content created by artificial intelligence applications, INFORMIT (Aug 1, 2027). 
• Lee-Ann Tong, Authorship of computer programs under South African copyright law:  Haupt t/a Soft Copy v Brewers Marketing Intelligence (Pty) Ltd: notes, 122 SALJ 513  (2005). 
• Nicola Lucchi, ChatGPT: A Case Study on Copyright Challenges for Generative  Artificial Intelligence Systems, 15 EJRR 602 (2024). 
• Paolo Guarda & Laura Trevisanello, Robots as artists, robots as inventors: Is the  intellectual property rights world ready? INFORMIT (Nov 1, 2021). 
• Sizwe Snail Ka Mtuze & Masego Morige, Towards drafting artificial intelligence (AI)  legislation in South Africa, 45 SAJ 162 (2024). 

Treaties 

• Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) (15 April  1994, entered into force 1 January 1995) 1869 UNTS 299. 

Internet Sources 

• Anthony Leung, Creative Machines? The Riddle of AI and Copyright Authorship and  Ownership, LEXOLOGY (Jun 2, 2023) https://www.lexology.com/library/detail.aspx?g=92277358-605b-43fc-b8b0- 93db1707f37d 
• Anonymous, Copyright office issues new guidance on AI-assisted works, but legal  concerns linger, EVERSHEDS SUTHERLAND (Apr 21, 2023) https://www.eversheds-sutherland.com/en/south-africa/insights/copyright-office issues-new-guidance-on-ai-assisted-works-but-legal-concerns-linger
• Cliffe Dekker Hofmeyr, AI ownership and machines as creators, CHD (Jul 8, 2029) https://www.cliffedekkerhofmeyr.com/en/news/publications/2019/technology/technol ogy-alert-8-july-2019-ai-ownership-and-machines-as-creators.html 
• Iphigeniefossati, Who owns AI-generated content? Copyright laws explained (2025),  TECHLAWSPHERE LIMITED, (Mar 31, 2025) https://techlawsphere.com/2025/03/31/who-owns-ai-generated-content-copyright laws-explained-2025/
• John S. McKeown, Appendix B40. A Consultation on a Modern Copyright Framework  for Artificial Intelligence and the Internet of Things, WESTLAW https://intl.westlaw.com/Document/Iee420312e5ce11ecae23a11e97e51250/View/Full Text.html 
• Julia Dickenson & Alex Morgan, et al, Creative machines: Ownership of copyright in  content created by artificial intelligence applications, INFORMIT (Aug 1,2027)  https://search.informit.org/doi/10.3316/agispt.20230206082916 . 
• Neha Tripathi, Generative AI and Copyright Ownership: Who Owns AI-Created  Works?, THE LAW COMMUNICANTS (Jun 19, 2025) https://thelawcommunicants.com/generative-ai-and-copyright-ownership-who-owns ai-created-works/ 
• Nicola LUCCHI & Serra Hunter, et al, Generative AI and Copyright, EUROPEAN  PARLIAMENT ( Jul,2025)  https://www.europarl.europa.eu/RegData/etudes/STUD/2025/774095/IUST_STU(202 5)774095_EN.pdf#:~:text=This%20study%20examines%20how%20generative%20A I%20challenges%20core,exceptions%2C%20and%20the%20uncertain%20status%20 of%20AI-generated%20content. 
• Paolo Guarda & Laura Trevisanello, Robots as artists, robots as inventors: Is the  intellectual property rights world ready? INFORMIT (Nov 1,2021)  https://search.informit.org/doi/10.3316/agispt.20221211079608

1 Neha Tripathi, Generative AI and Copyright Ownership: Who Owns AI-Created Works?, THE LAW  COMMUNICANTS, (Jun 19,2025) https://thelawcommunicants.com/generative-ai-and-copyright-ownership who-owns-ai-created-works/. 

2Ibid. 

3 BARRY B SOOKMAN COMPUTER, INTERNET, AND ELECTRONIC COMMERCE LAW (e-book) 9 (1st ed,1989).  https://intl.westlaw.com/Document/Iaebc74c3421711ecb979c967a094f0c2/View/FullText.html?ppcid=4eb1de47 444a42899ff2379480a3817c&originationContext=documenttoc&transitionType=CategoryPageItem&contextDa ta=%28sc.Default%29 

4 Enrico Bonadio & Luke McDonagh, Artificial intelligence as producer and consumer of copyright  works: evaluating the consequences of algorithmic creativity, 2 IPQ 1, 1 (2020). 

5 Nicola Lucchi, ChatGPT: A Case Study on Copyright Challenges for Generative Artificial Intelligence Systems,  15 EJRR 602, 608 (2024).

6 Paolo Guarda & Laura Trevisanello, Robots as artists, robots as inventors: Is the intellectual property rights  world ready? INFORMIT (Nov 1,2021) https://search.informit.org/doi/10.3316/agispt.20221211079608

7Julia Dickenson &Alex Morgan, et al, Creative machines: Ownership of copyright in content created by artificial  intelligence applications, INFORMIT (Aug 1,2027)  https://search.informit.org/doi/10.3316/agispt.20230206082916. 

8 Nicola LUCCHI & Serra Hunter, et al, Generative AI and Copyright, EUROPEAN PARLIAMENT ( Jul,2025) https://www.europarl.europa.eu/RegData/etudes/STUD/2025/774095/IUST_STU(2025)774095_EN.pdf#:~:text=This%20study%20examines%20how%20generative%20AI%20challenges%20core,exceptions%2C%20and% 20the%20uncertain%20status%20of%20AI-generated%20content. 

9Ibid. 

10 Anthony Leung, Creative Machines? The Riddle of AI and Copyright Authorship and Ownership, LEXOLOGY (Jun 2, 2023) https://www.lexology.com/library/detail.aspx?g=92277358-605b-43fc-b8b0-93db1707f37d. 

11 Copyright, Designs and Patents Act,1988, S 9(3). 

12 BARRY B SOOKMAN COMPUTER, INTERNET, AND ELECTRONIC COMMERCE LAW (e-book) 24 (1st ed,1989). https://intl.westlaw.com/Document/Iaebc74c3421711ecb979c967a094f0c2/View/FullText.html?ppcid=4eb1de47 444a42899ff2379480a3817c&originationContext=documenttoc&transitionType=CategoryPageItem&contextDa ta=%28sc.Default%29.

13 The Copyright Act 98 of 1978, S 2. 

14 Cliffe Dekker Hofmeyr, AI ownership and machines as creators, CHD (Jul 8,2029)  https://www.cliffedekkerhofmeyr.com/en/news/publications/2019/technology/technology-alert-8-july-2019-ai ownership-and-machines-as-creators.html. 

15 Lee-Ann Tong, Authorship of computer programs under South African copyright law : Haupt t/a Soft  Copy v Brewers Marketing Intelligence (Pty) Ltd : notes, 122 SALJ 513, 516-517 (2005). 16 Ibid.

17 Thaler v Perlmutter No 22-1564, (2023) ( United of State). 

18 Ibid. 

19 Ibid. 

20 Anshul Kumaria & MIT World Peace University, The copyright issue in ai-generated content:  legal challenges and future directions, 6 IJL & LR 1765, 1767 (2024). 

21 BARRY B SOOKMAN COMPUTER, INTERNET, AND ELECTRONIC COMMERCE LAW (e-book) 17 (1st ed,1989).  https://intl.westlaw.com/Document/Iaebc74c3421711ecb979c967a094f0c2/View/FullText.html?ppcid=4eb1de47 444a42899ff2379480a3817c&originationContext=documenttoc&transitionType=CategoryPageItem&contextDa ta=%28sc.Default%29. 

22 Payen Components SA Ltd v Bovic CC and Others (1995) (4) SA 441 (A) (South Africa). 23 1995 (4) SA p450. 

24 1995 (4) SA p449 and 1995 (4) SA p450.

25 1995 (4) SA p449. 

26 Haupt v Brewers Marketing Intelligence (Pty) Ltd (2006) SCA 39 (South Africa). 

27 Copyright Act 98 of 1978,S 1(i). 

28 Haupt v Brewers Marketing Intelligence (Pty) Ltd (2006) SCA 39 (South Africa). See: Lee-Ann Tong,  Authorship of computer programs under South African copyright law : Haupt t/a Soft Copy v Brewers Marketing  Intelligence (Pty) Ltd : notes, 122 SALJ 513, 525 (2005). 

29 Lee-Ann Tong, Authorship of computer programs under South African copyright law : Haupt t/a Soft Copy v  Brewers Marketing Intelligence (Pty) Ltd : notes, 122 SALJ 513, 522-523 (2005). 

30 Ibid. 

31 Nicola Lucchi, ChatGPT: A Case Study on Copyright Challenges for Generative Artificial Intelligence Systems,  15 EJRR 602, 609 (2024). 

32 BARRY B SOOKMAN COMPUTER, INTERNET, AND ELECTRONIC COMMERCE LAW (e-book) 17 (1st ed,1989).  https://intl.westlaw.com/Document/Iaebc74c3421711ecb979c967a094f0c2/View/FullText.html?ppcid=4eb1de47 444a42899ff2379480a3817c&originationContext=documenttoc&transitionType=CategoryPageItem&contextDa ta=%28sc.Default%29.

33 Payen Components SA Ltd v Bovic CC and Others (1995) (4) SA 441 (A) p449. 

34 BARRY B SOOKMAN COMPUTER, INTERNET, AND ELECTRONIC COMMERCE LAW (e-book) 20-21 (1st ed,1989). https://intl.westlaw.com/Document/Iaebc74c3421711ecb979c967a094f0c2/View/FullText.html?ppcid=4eb1de47 444a42899ff2379480a3817c&originationContext=documenttoc&transitionType=CategoryPageItem&contextDa ta=%28sc.Default%29. 

35 Nicola Lucchi, ChatGPT: A Case Study on Copyright Challenges for Generative Artificial Intelligence Systems,  15 EJRR 602, 609-610 (2024).

36 Neha Tripathi, Generative AI and Copyright Ownership: Who Owns AI-Created Works?, THE LAW  COMMUNICANTS, (Jun 19,2025) https://thelawcommunicants.com/generative-ai-and-copyright-ownership who-owns-ai-created-works/. 

37 Anshul Kumaria & MIT World Peace University, The copyright issue in ai-generated content:  legal challenges and future directions, 6 IJL & LR 1765, 1766-1767 (2024). 

38 Ibid. 

39 Nicola Lucchi, ChatGPT: A Case Study on Copyright Challenges for Generative Artificial Intelligence Systems,  15 EJRR 602, 609-610 (2024). 

40 Anonymous, Copyright office issues new guidance on AI-assisted works, but legal concerns linger,  EVERSHEDS SUTHERLAND, (Apr 21,2023) https://www.eversheds-sutherland.com/en/south africa/insights/copyright-office-issues-new-guidance-on-ai-assisted-works-but-legal-concerns-linger

41 Article 4(3) of Directive (EU) 2019/790. 

42 João Pedro Quintais, Generative AI, copyright and the AI Act, 56 CL & SR 1, 5 (2025).

43 João Pedro Quintais, Generative AI, copyright and the AI Act, 56 CL & SR 1, 5-6 (2025).

44 Sizwe Snail Ka MtuzeI & Masego Morige, Towards drafting artificial intelligence (AI) legislation in South  Africa, 45 SAJ 162, 162 (2024).

45 Anonymous, Copyright office issues new guidance on AI-assisted works, but legal concerns linger,  EVERSHEDS SUTHERLAND, (Apr 21,2023) https://www.eversheds-sutherland.com/en/south africa/insights/copyright-office-issues-new-guidance-on-ai-assisted-works-but-legal-concerns-linger

46 Iphigeniefossati, Who owns AI-generated content? Copyright laws explained (2025), TECHLAWSPHERE  LIMITED,( Mar 31, 2025) https://techlawsphere.com/2025/03/31/who-owns-ai-generated-content-copyright laws-explained-2025/ 

47 Anonymous, Copyright office issues new guidance on AI-assisted works, but legal concerns linger,  EVERSHEDS SUTHERLAND, (Apr 21,2023) https://www.eversheds-sutherland.com/en/south africa/insights/copyright-office-issues-new-guidance-on-ai-assisted-works-but-legal-concerns-linger.

48 John S. McKeown, Appendix B40. A Consultation on a Modern Copyright Framework for Artificial Intelligence  and the Internet of Things , WESTLAW, https://intl.westlaw.com/Document/Iee420312e5ce11ecae23a11e97e51250/View/FullText.html?navigationPath= Search%2Fv1%2Fresults%2Fnavigation%2Fi0a89830f00000199783ed5383f2405f0%3Fppcid%3D60b1cc5d4b 9a442d85b06c0e1d4f8f0e%26Nav%3DINTERNATIONAL  ANALYTICAL%26navQualifier%3DI1decad30a08911e28578f7ccc38dcbee%26fragmentIdentifier%3DIee420 312e5ce11ecae23a11e97e51250%26parentRank%3D0%26startIndex%3D1%26contextData%3D%2528sc.Searc h%2529%26transitionType%3DSearchItem&listSource=Search&listPageSource=b7b3042af1595065106df1d12 2344a04&list=ALL&rank=1&sessionScopeId=a1d9ca3e0c9d16f9bd4de2cd0a4d98fcf6411015ec9057da9b6e9f 14f1af248d&ppcid=60b1cc5d4b9a442d85b06c0e1d4f8f0e&originationContext=Search%20Result&transitionTy pe=SearchItem&contextData=%28sc.Search%29. See: Enrico Bonadio & Luke McDonagh, Artificial intelligence  as producer and consumer of copyright works: evaluating the consequences of algorithmic creativity, 2 CRO 1,  33-34(2020). 

49 Enrico Bonadio & Luke McDonagh, Artificial intelligence as producer and consumer of copyright works:  evaluating the consequences of algorithmic creativity, 2 CRO 1, 34-35 (2020). 

50 John S. McKeown, Appendix B40. A Consultation on a Modern Copyright Framework for Artificial Intelligence  and the Internet of Things , WESTLAW, page 10  https://intl.westlaw.com/Document/Iee420312e5ce11ecae23a11e97e51250/View/FullText.html?navigationPath=  Search%2Fv1%2Fresults%2Fnavigation%2Fi0a89830f00000199783ed5383f2405f0%3Fppcid%3D60b1cc5d4b 9a442d85b06c0e1d4f8f0e%26Nav%3DINTERNATIONAL  ANALYTICAL%26navQualifier%3DI1decad30a08911e28578f7ccc38dcbee%26fragmentIdentifier%3DIee420 312e5ce11ecae23a11e97e51250%26parentRank%3D0%26startIndex%3D1%26contextData%3D%2528sc.Searc h%2529%26transitionType%3DSearchItem&listSource=Search&listPageSource=b7b3042af1595065106df1d12 2344a04&list=ALL&rank=1&sessionScopeId=a1d9ca3e0c9d16f9bd4de2cd0a4d98fcf6411015ec9057da9b6e9f 14f1af248d&ppcid=60b1cc5d4b9a442d85b06c0e1d4f8f0e&originationContext=Search%20Result&transitionTy pe=SearchItem&contextData=%28sc.Search%29. 

51 Ibid. 

52 Enrico Bonadio & Luke McDonagh, Artificial intelligence as producer and consumer of copyright works:  evaluating the consequences of algorithmic creativity, 2 CRO 1, 36 (2020). 

53 Neha Tripathi, Generative AI and Copyright Ownership: Who Owns AI-Created Works?, THE LAW  COMMUNICANTS, (Jun 19,2025) https://thelawcommunicants.com/generative-ai-and-copyright-ownership who-owns-ai-created-works/.