Artificial Intelligence and Copyright Protection

by Kanerva Jalas

Intellectual property (IP) is a fundamental topic in the legal field regarding the development of technology. Intellectual property rights refer to ownership over intangible goods, and can in general be differentiated to four main categories: trademarks, industrial designs, copyright and patent protection. Copyright focuses on the protection of creative works, such as music, paintings, writings or other visual creations. Additionally, most national jurisdictions include so-called related rights that can be granted to performing artists, phonograms and producers. The rationale for copyright of creative works relies on two main theories: first, it is seen that the intellectual creations of individuals are extensions of their personality, therefore justifying the existence of moral and economic rights. Second, granting rights to the author of a work is believed to work as an incentive and encouragement for individual creative work. In most countries, copyright protection exists where there is a creative work which falls under subject matter of protection. Such works include, for example, those falling under literary, scientific and artistic fields. Secondly, the creative work must be original, and it must consist of the author’s own expression of an idea. Third, this expression must fall under the scope of protection. In the European Union (EU), the author has several economic and moral rights, such as the right of reproduction, the right of communication and making the work available to the public, and the right of distribution.

The term ‘artificial intelligence’ (AI) includes a broad range of systems. The definition can differ depending on the subject-matter. In general, AI systems can be defined as systems which make use of logical reasoning and decision-making which are usually performed by humans. Most AI systems nowadays still consist of so-called ‘quasi-AI’ models, which cannot truly be independent from human input. The lack of the originality from such quasi systems results from the fact that the software cannot create anything novel, as it can only gather ‘knowledge’ from the database that has been fed to it by humans. Hence, such systems are often also called ‘AI-assisted’ models. However, recent developments have resulted in the creation of AI neural-network systems which are capable of copying the functioning of a human brain to such an extent where the outputs of the system can be declared as new and original. Neural network systems are built on the model of neurons in a human brain. Such functioning allows the system to learn from its mistakes, and hence build upon failed creations. Therefore, the AI system is able to learn to write and to generate copyright protectable content through ‘practise’. Differentiation between the quasi-AI and neural network systems can be understood through characteristics that allow the new variation to make creative works: first, neural networks do not merely gather data from sources which it has been fed. Instead, they operate by making calculated choices in introducing a new work. Second, developed AI is able to learn; it will process the information it receives and implements feedback in constant new attempts at its task. Third, it is able to function independently. Such a definition can be granted to a model which can function and understand tasks without any intervention from external parties. Finally, the system’s outputs consist of unpredictable results. This characteristic refers to the ability of the system to act as a ‘black box’, meaning that no one can fully be aware of the process which the system makes to arrive at its outputs.

The creative ability of such AI systems raise a fundamental question: can the works of AI be protected under copyright laws, and who is the author of such a creation? Due to the territorial nature of copyright legislation, each jurisdiction will have their specific definitions of what constitutes an author and what constitutes AI. However, due to globalisation, many countries have opted for similar lines of development in approaching the issue. In the European Union (EU), legislation such as the Term Directive, the Software Directive and the Database Directive have been enacted to address the situation. According to article 2 (1) of the Software Directive the author of a computer program is either the natural person or the group of natural persons who have developed the system. This line of reasoning has been followed in the case of the Court of Justice of the European Union (CJEU), where the Court announced that in order to be protected, the work should reflect the free and creative choices of the author, and that it must be their ‘own intellectual creation’. The legislation and the case law implies a requirement of human involvement in the creation of a work. Copyright legislation in the United States (US) takes a step even further: Section 313 of the Compendium of US Copyright Office Practises states that the subject-matter which falls under protection of copyright includes only those works where its author is a human being. The legislation therefore leaves a gap for the works produced by AI independently (also referred to as ‘AI-generated content’’).

The requirement of human involvement for copyright protected subject-matter flows from arguably from the concept of ownership. The consequences of granting property (in this case intellectual property) to a non-human entity would require such a system to be granted legal personality. Such a decision in turn would have large-scale consequences on our society as a whole: through legal personhood, AI systems would be able to derive rights and responsibilities also in other areas apart from copyright protection. However, such fear of consequences will leave the problem of content protection for AI-generated content unaddressed. In the combat of a legislative gap, some jurisdictions have been faster than others: in the United Kingdom (UK), India and Hong Kong, regulations have been adopted which solve the AI copyright issue by granting the rights for the creations developed by AI systems to the person who has undertaken ‘the arrangements necessary for the creation of the work’. Opinions on solutions have also been suggested by experts in the field. One such solution includes an idea that all AI-generated content would be in the public domain.

It will be left to be seen which approaches different national legislators choose to adopt, and whether they will decide to stick with the chosen option. What is certain, however, is that the development of independent AI-generated content will bring forth an important worldwide discussion related to the question of how we should treat AI, and whether we should grant them legal rights.

Sources:

Berne Convention for the Protection of Literary and Artistic Works (1886).

US Copyright Office, Compendium of US Copyright Office Practices (3d ed. 2021)

Directive 2001/29/EC of the European Parliament and of the Council of 22 May 2001 on the harmonisation of certain aspects of copyright and related rights in the information society [2001] OJ L167/10.

Case C-310/17 Levola Hengelo BV v Smilde Foods BV EU:C:2018:899.

Case C-145/10 Eva-Maria Painer v Standard VerlagsGmbH EU:C:2011:798.

Case C-606/10 Football Dataco v Yahoo UK and others EU:C:2012:115.

Case C-5/08 Infopaq International A/S v Danske Dagblades Forening EU:C:2009:465.

Annette Kur and others, European Intellectual Property Law: text, cases and materials (2nd ed, Edward Elgar Publishing, 2019).

S A Merrill and W J Raduchel (eds), Copyright in the Digital Era: Building Evidence for Policy (National Academies Press, 2013).

Shlomit Yanisky-Ravid, ‘Generating Rembrandt: Artificial Intelligence, Copyright and Accountability in the 3A Era: The Human-like Authors Are Already Here: A New Model’ (2017) Michigan State Law Review 659.

Andres Guadamuz, ‘Artificial Intelligence and Copyright’ (WIPO Magazine, October 2017): <https://www.wipo.int/wipo_magazine/en/2017/05/article_0003.html> accessed on the 22nd of January 2023.

Pratap Devarapalli, ‘Machine Learning to Machine Owning: Redefining the Copyright Ownership from the perspective of Australian, US, UK and EU Law’ (2018) 40 European Intellectual Property Review 722.