Keywords

1 Introduction

The fourth industrial revolution and Web 3.0 are no longer a matter of fiction and phantasy. In this new reality, where the borders between the physical and the digital world are not as clear as they used to be, everyone is talking about blockchain,Footnote 1 NFTs,Footnote 2 Artificial Intelligence (hereinafter: AI)Footnote 3 and the metaverse.Footnote 4 Whilst this era is exciting, transformative, and offering a vast of new opportunities in almost every segment of our private and work lives, the technological progress is at the same time also shifting paradigms and challenging several “real world”-concepts, in particular in the sphere of legal regulation. The legislator is traditionally known for being a “late bloomer” when it comes to adapting to technological advancements. Nonetheless, the speed at which the current progress is moving in almost every field of industry is, to be frank, not very easy to catch up with, even for the ones that have their hand strongly placed on the pulse of the time.

One of the fields of law particularly challenged by these developments is Intellectual Property (hereinafter: “IP”). Throughout history, IP and technological growth have always been closely intertwined. Then the progress of the former has been a catalyst for introducing a comprehensive regulatory framework for the latter, to acknowledge and incentivize the creators and to further promote technological advancements and the dissemination of knowledge. At the same time, rapid technological progress—in particular the digitization—also constantly challenged IP, even to the point when the justification of its existence was questioned. However, this field of law has shown immense “shape shifter” qualities and managed to persevere, legislatively transform, and adapt to new technological circumstances.

With the emergence of AI-systems however, the world of IP was turned upside down. Perhaps even more severely than ever before. The legal field dedicated to protection of creations of the human mind was abruptly confronted with the dilemma of the legal status and legal protection of output of the simplified speaking, mechanical or artificial minds/machines. In this new era, the main question is not any more how to stop unlawful (digital) reproduction of human intellectual creations, but whether they can be replaced by the creations of the machines.Footnote 5 Some of the legal categories of IP that were considered rather self-explanatory for centuries, now potentially need to be reassessed, re-evaluated and perhaps even thoroughly redefined in the light of the “power of AI”. The latter include, but are not limited to: the concept of the author, the concept of the inventor, as well as the (human) creator in general, the notion of originality of a work of art, the criteria for assessing the “inventive step” or “state of the art”, the legal fiction of the “average expert” etc.

Among several questions raised by AI in the field of IP, this chapter will provide a short overview of the general impact of AI-systems on the principles of copyright and patent law, before investigating the challenge to the concept of authorship in particular.

2 AI—Challenges for Patent and Copyright Law

Although discussions have become very intense in the last years, the fact remains that the potential impact of AI on the IP has already been addressed in the early 1990s by the World Intellectual Property Organization (WIPO).Footnote 6

So far, there is no clear and generally accepted definition of it.Footnote 7 For the purpose of this paper, the one we choose is that AI represents a branch of computer science, often described as computer—based systems, that are developed to imitate human behavior.Footnote 8 In other words, AI-systems perform cognitive tasks that are inherent to humans, and they include a. o. learning, analysis, reasoning and problem solving. At the core of AI systems are software and data.Footnote 9 Hence, AI is based on computational models and algorithms, which themselves are built on the rules provided by the software applied to the data that is used as an input. The simulation and imitation of intelligent behavior by the AI can also result in outputs in art, literature, and science, as well as in a number of technical fields. If those outputs had been a result of human intellectual or creative work, they could potentially be protected by copyright, or a patent. Their creators would have the status of an author, or an inventor.

Questions have been raised about the algorithmic “creations”, or outputs of the AI. Can AI be an author? Can AI be an inventor? Or do we have “creations without a creator”?Footnote 10

The fundaments of copyright and patent law and the justification for the protection of creations under this regime are inherently anthropocentric. These rights are granted to (human) creators to acknowledge and honor their intellectual accomplishments on the one hand. On the other, they are granted as incentive for those creators to continue their work and promote creativity and technological and societal development in the public interest. Consequently, IP is devoted to the human element and human contribution in the process of creation.

However, a fact is—AI “creates”, but one could also say it solely produces. Although imitation (in this case of human cognitive processes) perhaps does represent the greatest form of flattery, this “computational admiration” can have negative effects as well. What happens when the AI-outputs and human creations start competing in the field of art and industry? What if end-users do not recognize the difference between the AI-product and human works of art and inventions, which can already be the case (comparable to Turing Test)Footnote 11? What if the human creator, or the human contributor to the process of AI-systems generating output, becomes obsolete? Is the whole system truly so drastically at stake?

The European Union (hereinafter: “EU”) has taken a stand in this regard by the Resolution of the European Parliament on IP and AI, proclaiming the following: “…a human-centered approach to AI that is compliant with ethical principles and human rights is needed, if the technology is to remain a tool that serves people and the common good.”Footnote 12 This approach is in line with the Communication from the European Commission on AI for Europe, which states that “focus on investing in people is a cornerstone of a human-centric, inclusive approach to AI, and will require a significant investment”, as well as that “[i]ndividuals should be able to control the data generated by using these tools and should know whether they are communicating with a machine or another human”.Footnote 13

Hence, it seems like the human creator remains the pivot of IP protection also in the future. However, the technology is advancing, and we cannot stop it. This means that AI-systems themselves and/or their outputs, if there is sufficient justification and a potential market failure—which this paper will not examine—need to be awarded some type of protection. Can the current IP laws and conventions be the legal instrument for such protection?

When it comes to IP-protection of AI-generated output in general, there are voices for and against it. Some of the arguments used in the course of this discussion opposing the protection are, that it would devaluate human creative efforts.Footnote 14 The opinions in favour state that protection would be necessary for the purpose of stimulating investment for development and introduction of AI and that it is questionable whether it would be enough to incentivise the human developers of AI by giving them an IP right on the AI itself, without awarding them at the same time the right to exclusively use the output of that given AI.Footnote 15

In case the protection would be awarded, it remains unclear, who would be its be beneficiary?

When there is a satisfactory scope and quality of human creative contribution to the output generated by the AI, in the sense that AI was used as a tool for the human creative process (“AI-assisted creations”), the current IP framework with its human-centric approach remains applicable.Footnote 16

But how much direct streamlining and guiding of the creative process from the human side would be enough? The sole development of the algorithm would obviously not suffice.Footnote 17 What could potentially represent an acceptable amount of creative choices of an individual would be the selection of training data, the definition of the goal of the process, and the choice of the learning protocol based on those data, as well as the possibility to modify the outcome developed by the AI by adding a title, or a commentary on its purpose.Footnote 18

It gets challenging when AI agents act potentially completely autonomously (so-called “strong AI”Footnote 19 or “Artificial Super-Intelligence”Footnote 20), one might even say “creative”, and the human piloting of the process to a predictable output is being eliminated. This could be the case with algorithms based on machine-learning and deep learning processes.Footnote 21

However, as several authors stress,Footnote 22 according to the current state of art, the sovereignty of the existing AI-technologies is overestimated. AI systems are in no way fully autonomous, and their outputs are predictable. Human control and human will over the frame/scope of AIs actions still exist—it is a human decision how autonomous an AI can get (so-called “weak AI”). Nevertheless, that does not prevent owners, developers or users of AI, who are (paradoxically) humans, from crediting AI-systems with authorship of various works of art,Footnote 23 or as inventors of technical innovations (e.g. DABUS),Footnote 24 both representing outputs developed by using those systems. Hence, a clear distinction needs to be made between what the technology is independently capable of on one hand and the attributes awarded to the latter on the other. The result of this differentiation gives us also some direction as to whether the current patent and copyright law is sufficient to handle the AI-challenge, or it needs to be amended.

3 Copyright Law—The Concept of Authorship and AI

3.1 Nature of AI—Outputs

Art in its broadest sense is very powerful. It can shape societies and the public opinion,Footnote 25 it can provoke, challenge the existing regimes and outdated ways of thinking, it can evoke the complete spectrum of emotions. Why? Because it is created by humans. And no matter how philosophical or ethical this argument may seem in this legal discussion; it does carry weight. AI can imitate cognitive functions of the human brain (analysis, recognizing patterns, drawing conclusions, …), nevertheless it is questionable, if it can also imitate human creativity and the human “spirit”. The individual stamp of the author on the work of art he/she is creating is also a result of other brain functions such as e.g. imagination, expression of various emotions (affection, fear, anger, sorrow, …), and others.Footnote 26 In addition to that, the sources of creative action are, among others, personality traits and biographical experiences.Footnote 27 These are all categories that an AI system cannot possess. Finally, the creative choices a human author makes in the process of creating a work are also ruled by motives, and those vastly differ with regard to humans and AI.Footnote 28 That is, if we can even speak of motives in the production process (theoretically solely) generated by the latter.

Hence, it seems that AI systems produce creations that only “look like literary and artistic works”,Footnote 29 and that the law should not “reward trickery” (in this context, with crediting authorship to AI). Although the outputs of AI systems are very close to what we recognise as works of art, they are just passing off as ones.Footnote 30 There is no work of art without a creative choice of a human author.

There is quite an impressive catalogue of these “lookalikes” in the portfolio of various AI systems.Footnote 31 This includes “The Next Rembrandt”, a “painting” generated by an AI that had analysed thousands of works of Rembrandt Harmenszoon van Rijn, or the “Portrait of Edmond de Belamy,” which is the first AI “artwork” sold at Christie’s.Footnote 32 Additionally, there is the first AI musical album “Hello World” by SKYGGE.Footnote 33

But are they “lookalikes” or truly works of art, since, as stated above, the technological progress has apparently not yet reached the state at which AI-systems can act as fully autonomous agents. The algorithms and systems still depend on human guidance. If we look closer into the projects and the background of the above-named AI “works”, we actually see this human navigation clearly. Then behind the “Portrait of Edmond de Belamy,” is a collective of researchers and artists called “Obvious”, who work with the latest models of deep learning to explore “the creative potential of AI”. The situation is very similar with SKYGGE. Of course, this will not always be the case, but when the human author explores and uses advantages of AI as a tool to assist him/her in the process of creation, the authorship over the AI-assisted output belongs to that person.Footnote 34

Nevertheless, to which extent and scope human involvement in the process of creation is required in order to consider AI only as a tool of the creator, is eventually hard to determine.Footnote 35 There is no definite answer, since it would be ambitious to set a general threshold for establishing a causal chain between the natural person and the AI output, sufficiently strong to justify awarding a person the status of author. Then the latter is a very case-dependent situation.Footnote 36

Another question is, exactly which one of the human actors (or number of them) involved in the process of AI “creation” would be awarded authorship in that case? There are three key categories of persons, whose contributions are of importance for an AI-system: the owner (investor), the developer (or a team of developers) and the user (trainer).Footnote 37 An idea would be to give it to the developer of the AI, but another person could have worked on the design of the AI, a third one could have provided the AI with training data etc.Footnote 38

In any case natural persons are the ones making creative choices.Footnote 39 They are the ones in the root of the requirement for a work of art to be an individual creation of the mind, and for awarding such creations copyright protection.

3.2 Human-Centric Copyright

The anthropocentric approach to copyright is also to a large extent the position represented in the acquis communautaire and the copyright legislation of the EU Member States,Footnote 40 as well as the case law of the Court of Justice of the European Union (hereinafter: “CJEU”)Footnote 41 when referring to the notions of “work”, “originality”, “personality”, “personal touch” etc. For instance, in its Smilde judgement,Footnote 42 the CJEU stressed that there are two cumulative conditions, which must be satisfied for subject matter to be classified as a “work” (within the meaning of Directive 2001/29). These are that the subject matter concerned must be original in the sense that it is the author’s own intellectual creation, and that only something which is the expression of the author’s own intellectual creation may be classified as “work” within the meaning of the directive in question. Furthermore, the CJEU pointed out that copyright within the meaning of Article 2(a) of Directive 2001/29 is liable to apply only in relation to a subject-matter which is original in the sense that it is the own intellectual creation of the author.Footnote 43 Finally, it identified that the latter is an intellectual creation of the author reflecting his personality and expressing his free and creative choices in the production of that work,Footnote 44 and that by making those various choices, the author [of a portrait photograph] can stamp the work created with his “personal touch”.Footnote 45

Those are all qualities of natural persons. No machines in sight. Apart from this choice of a humanistic approach to the concept of authorship, there is a further reason why this status cannot be awarded to an AI-system: the lack of legal personhood.

To be an author is a privilege, but also a responsibility, since he/she can be held liable for unauthorised use of someone’s copyright in their work, or their work of art can infringe some other rights (e.g. personality rights).Footnote 46 The AI, due to the lack of legal personhood, cannot.

A potential solution for that would be to introduce “E-personhood” for AI-systems. However, the European Parliament clearly declared in 2020 that it would not be appropriate to seek to impart legal personality to AI technologies and pointed out the negative impact of such a possibility on incentives for human creators.Footnote 47

However, there are also legislations that do not necessarily invariably follow the principles of the “human-centric” Droit d’ Auteur—approach, which is aiming toward protecting the human creator and his interests. They have a more investment-oriented approach to copyright. For example, Art. 9 (Authorship of work), sec. 3 of the UK Copyright, Designs and Patents Act 1988 states that: “In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken.” But not only the UK, also South Africa, Hong Kong, India, Ireland and New Zealand have comparable legislative solutions, which would make it possible to grant protection for an e.g. computer-generated work.Footnote 48 Hence, it is possible that at some point we experience a division by countries in the legislative attitude toward the question, who can be considered an author of an AI-assisted work.

But this still does not mean that the authorship would be awarded to an AI-system. It would be credited to a person, who in this case could also be a legal person. The answer to the question who exactly that person is—“by whom the arrangements necessary for the creation of the work are undertaken”—in the case of an AI-assisted works remains unclear. It could be the developer, the owner of the program, the person who controls the AI, or who feeds it with necessary data.Footnote 49

4 Final Remarks

The scope and substance of the changes that AI will bring in the field of IP rights are still uncertain. Some authors hold the view that these amendments will inevitably occur, when the technology reaches the point of development, that AI-systems can make (independent) choicesFootnote 50 (“strong AI”). This opinion can be supported, because when that occurs, the current lawmaker’s approach might not be sufficient to achieve legal certainty in this field. Nonetheless, perhaps we should not wait that long and begin with cautiously adapting the current legislative framework. Then already the existing “weak AI”, which assists the human creators and serves as their tool, poses a number of questions and generates some legal uncertainties. These include the sufficient scope and nature of the human involvement in the process of creation and granting a person the authorship over the AI-assisted work of art, as well as the criteria for defining who that person is (developer, user, investor, …).

Although it always seems that there is quite some urgency to regulate when a new technology appears, it is usually advisable not to (legislatively or otherwise) overeact.Footnote 51 Precisely this has been “the tactics” of IP, as a highly tech-susceptible field of law, for centuries. With a lot of success, if we may notice. The majority of core principles of e.g. copyright and patent law have remained largely unchanged throughout technological storms, which therefore seems to be a reasonable result when facing the AI-challenges as well.

The regulatory goal would most certainly be to create a balance between IP and AI,Footnote 52 since the latter is obviously here to stay, but with a human-centred approach in mind, as the European Parliament proclaimed. Also, the latter announced that the new regulatory framework in the field of AI technologies and IP on the EU-level should take the form of a regulation and not a directive.Footnote 53 This would facilitate establishing equal standards across the Union, which some authors interpret as potentially announcing the adoption of a “General Copyright Regulation”.Footnote 54

When it comes to copyright and the notion of authorship, we need to admit that some of the characteristics of the system are to an extent antiquated in today’s time and seem even a bit romantic.Footnote 55 Perhaps the strong personal tie between the author and his creation is no longer existent for all types of works, partially due to the technical nature of some (e.g. computer programs), or the use of technical tools to create them (e.g. software or randomisers).Footnote 56 Nevertheless, this does not in any way justify a complete change of core concepts of copyright law, such as the anthropocentric approach to authorship. Then the machines produce. But the creations emanate from the human spirit. Or in the words of George Bernard Shaw: “You use a glass mirror to see your face. You use works of art to see your soul.”