Ours is a world that is constantly evolving. From the days of the photostat machine and the black-and-white camera to the present time marked by amazing breakthroughs like 3D Printing, virtual surgeries mediated by 5G connectivity and artificially intelligent robots, it is undoubtedly the case that technology has impacted man’s standard of living. And as the technological landscape continues to expand, new innovations are constantly introduced which quickly integrate themselves into mainstream society. A good example would be the recent launch of ChatGPT- a natural language processing tool driven by AI technology which is able to simulate human-like conversations, answer questions, perform research, compose musicals, and even write lines of code. Despite being launched only in November 2022, this impressive tool has received widespread acceptance in society and broken the record for being the fastest application to reach over 100million users.
This essay focuses on Artificial Intelligence (AI), its myriad application over the years, its promise and its possible constraints. AI has been applied to many sectors including healthcare, manufacturing, customer service, and marketing, with impressive success. It has also been applied to musical compositions equivalent in quality to those of legends like Mozart and Beethoven, to writing award-winning literary novels and just recently, by Google-owned AI company, Deep Mind, to generate music by itself after listening to sample recordings.
And this is where the pivotal question of Intellectual Property sneaks in. This is because IP is meant to protect copyrights of artistic, literary or musical works. It also protects related patents, trademarks, designs, and trade secrets. However, a cursory glance at some of the IP legislations and statutes around the world reveal that one of the major criteria for such protection is authorship or inventorship. And authorship usually connotes the presence of a “human” author. This has therefore led to some controversy around AI-generated works, particularly on the role of human input as a criteria for protection.
Flowing from this, this essay delves into the intersection between AI and IP, the criteria of human authorship before protection, the challenges surrounding this subject matter, and futuristic recommendations gleaned from present case studies.
Just as the name connotes, Intellectual Property encompasses the properties and creations of the intellect or mind. This definition was duly captured by the World Intellectual Property Organization (WIPO), where it defined IP as “creations of the mind – everything ranging from works of art to inventions, computer programs, trademarks and other commercial signs.”
The 5 major classes of IP widely recognized around the world today include:
- Copyrights, which protect literary and artistic works, and in recent times, even computer programs, databases, and software codes.
- Patents, which protect inventions and innovations.
- Trademarks, which are meant to protect distinctive signs, marks, logos and symbols used in business.
- Industrial Designs, which are meant to protect the aesthetic, ornamental or functionally designed aspects of products.
- Trade Secrets, which protect commercially sensitive information and know-how.
Due to the critical role intellectual property (IP) plays in fostering commerce, trade, and innovation, various measures (both nationally and internationally) have been taken to protect it On the international scene for instance, we have the Berne Convention for the Protection of Literary and Artistic Works, the Patent Cooperation Treaty, the Madrid Agreement Concerning the International Registration of Marks, the WIPO Copyright Treaty (WCT), amongst various others. Meanwhile, on the local scene, we have the Nigerian Copyright Act of 2022, Patents and Designs Act and the Trademarks Act. There is also in existence the IP Registry and the Nigerian Copyright Commission which oversee administering and enforcing these laws and performing regulatory compliance functions.
Up until recently, the figurative waters of IP have remained relatively calm, that is until the inception of this present era characterized by disruptive innovations stemming from AI. Given the significant role of AI in all these disruptive technologies such as nanotechnology, Big Data, robotics, and quantum computing, it is the case that AI now presents the trending topic for debate within the current IP framework.
AI means ‘Artificial Intelligence’ and it connotes the simulation of ordinary human intelligence by artificial machines or computers. The phrase was coined by American computer scientist, John McCarthy, in an article titled “Ascribing Mental Qualities to Machines” in 1979. AI has also been defined as a sentient super-program with the potential to evolve into a highly advanced program with extraordinary cognitive abilities, consciousness, moral understanding, and other remarkable qualities.
Since 1979, the field of AI has morphed multi-dimensionally, blossoming into various uses spanning areas such as genetic programming, automation, facial recognition, Internet of Things (IoT) and quantum computing. As a matter of fact, it has grown to a point where AI can now create other AIs. In 2018 for instance, Google built an AI (named AutoML) whose purpose is to make other AI which have been referred to as “Child AIs”. The concept of Child AIs has received some attention and was just recently defined in the following words “A Child AI is an algorithm model whereby a parent AI creates another artificial intelligence algorithm – the “Child AI”. This would imply that Artificial Intelligence has now learned how to “procreate” itself, by writing code and determining the parameters needed for the new algorithm to operate”
The Intersection between AI and IP
As AI continues on this path of expansion across various fields of endeavor, it has brought itself into contact with Intellectual Property. This is illustrated by the following examples. In 2016, a group of researchers unveiled a portrait titled The Next Rembrandt, an AI-generated artwork that had analyzed thousands of works by the 17th-century Dutch artist Rembrandt van Rijn and had used that data to replicate a modern-day rendition of his works. In 2018, another painting created by an AI algorithm was sold at an auction for over $400,000. Still in 2018, an AI connectionist machine was able to invent 2 devices for use within homes. One important fact to note is that these works/products generated by AI are very realistic and one cannot tell whether they have been created by a human or an AI.
AI has also served some useful purposes in Intellectual Property, such as AI algorithms being deployed to identify and enforce trademarks, scanning digital platforms for potential trademark infringements. This application of AI in trademark protection demonstrates the inexorable integration of technology in IP.
Likewise, AI’s foray into the field of IP has brought it into contact with databases, which are often protected under IP as copyrighted works based on their unique selection and arrangement. And AI’s ability to process and examine vast amounts of databases for various purposes also intersects with the field of privacy and data protection regulations. As such, striking a balance between utilizing AI for data-driven insights while respecting privacy rights is another question for determination in IP and legal frameworks.
The Issue of “Human” Authorship
Generally speaking, there are 2 elements that ought to be fulfilled before IP protection is bestowed. Firstly, it must be shown that the work was novel, not copied or unoriginal. Secondly, it must be shown that the work was someone’s brainchild, the creation of someone’s intellect, and application of creative efforts. While the former requirement may be fulfilled by AI systems in certain cases, the latter cannot. Machines can create independent works which would be considered new, however, the second requirement of “an author’s intellectual creation” presents an issue because the word “author” is usually taken as inherently linked to a human person.
Thus, the emerging issue in the intersection between artificial intelligence (AI) and intellectual property (IP) is the question of whether works produced exclusively by AI can be granted IP protection. This problem specifically arises in the contexts where the invention is autonomously generated by AI without any human inventor assisting in such production as it were, such as in the case of “Child AIs” described above. It is this particular aspect that presents significant hurdles for the existing IP framework.
On the one hand, it has been argued that insofar as AI technologies are merely tools deployed by humans to address a particular problem, the integrity of the IP system remains intact in terms of determining the rightful attribution of authorship. It is the individual who contributes to the intelligent and imaginative conception of an invention that is recognized as the inventor. Also, it is the person who exercises independent and creative choices in producing a copyrighted work that is acknowledged as the author, and this principle extends to even when the human inventor employs technological tools which themselves possess self-organizing capabilities.
On the other hand, opponents question the extent of human creativity involved and argue that AI’s capacity to generate content independently should be recognized, thereby challenging traditional notions which form the very core of IP ownership. It has been argued that for general-purpose AIs, such as Google’s Deep Dream Generator, the human user’s role is often limited to initiating the process through a simple button activation, after which it is the AI that carries out the entire creative process. Under this heading, it has been argued that the human user does not actively make the creative choices which result in the given output.
To analyze the component units and issues that arise in both stances, there is need to make reference to extant legislation as well as decided case law.
For Copyright, the Berne Convention for the Protection of Literary and Artistic Works does not expressly contain the definition of an author, however, a cursory glance at the text and the couching of the legislative language adopted indicate that an “author” is most likely a natural person. For patents, Rule 19(1) of the European Patent Convention connotes a human inventor by virtue of its requirements which include the author’s family name, given name and other requirements clearly not applicable to machines.
Judicial decisions in the EU also seem to follow this path of reasoning. In the case of Eva-Maria Painer v Standard VerlagsGmbH and Others, which came before the European Court of Justice, the decision of AG Trstenjak is to the effect that the author of a copyright work is the individual who exercises independent and creative choices, thereby expressing their unique personality through the work. This principle applies to all human creations, including those that involve the use of technological aids. Therefore, under current EU law, AI systems are not recognized as authors of works, as they lack the capacity for independent creative decision-making.
This position also received judicial flavor in even more recent cases such as Cofemel — Sociedade de Vestuário SA v G-Star Raw CV, where the same court held that there are majorly two cumulative requirements to qualify as a work: “Existence of an original object and expression of the author’s intellectual creation”. The second requirement was further explained as a representation or expression of “an intellectual creation” of a human author.
In several common law jurisdictions, the core position is still the same – which is that AI cannot author works – even though there are minor adjustments in the language of the law. For instance, Section 9.3 of the UK’s CDPA 1988provides that in situations where a human author cannot exactly be identified, such authorship is instead to be vested in some person “by whom the arrangements necessary for the creation of the work are undertaken”.
This provision is likewise present in Section 108 of the Nigerian Copyright Act, Section 5(2) (c) of the Copyright Act 1994 of New Zealand as well as Article 21 of the Irish Copyright and Related Rights Act 2000. It is claimed that this phraseology was included in order to make sure that one way or another, credit goes to an actual human author for the creation of invention of new products.
Also, before AI can even attempt to claim authorship of works, it ought to be the case they possess some sort of legal personality which may sue or be sued in the event of infringement. But this is currently not the case, and vesting legal personality on AI and machines generally would require a massive change to the law that goes fundamentally beyond IP law and extends into other subject areas such as corporate law, criminal law, etc., and even to other disciplines apart from law such as ethics, morals and philosophy.
Case study – The DABUS Case
The issue of IP rights being vested in non-humans is one that has been around for quite some time. Consider the case of Naruto v. Slater, in which a dispute arose as to whether a monkey could claim rights over a selfie it had taken on a photographer’s camera. Of recent, there has also been another interesting case study on the issue of registration of non-human generated works in the DABUS case.
Towards the end of 2018, Dr. Stephen Thaler submitted two European patent applications – EP 18 275.163 (which was for a food container) and EP 18 275.174 (which was for devices and methods to attract enhanced attention). Dr. Thaler then filed a separate designation indicating the AI-based machine named “DABUS” as the inventor for both applications. He contended that because DABUS had the ability to “recognize the uniqueness of its own concepts before any human being,” the machine should be acknowledged as the creator, while he would assume the role of DABUS’ employer.
However, the Legal Board of Appeal of the EPO rejected this reasoning, and in its detailed judgment, stated that the applicant’s designation of inventorship did not meet the requirements set out in Article 81 and in Rule 19 (as mentioned earlier) of the European Patent Convention which provide that designations of inventorship must contain “a family name, given names and full address of the inventor who must be a natural person”. Irrespective of the fact that other requirements such as novelty and technical application were met, the inventions could not be patented because they specified a non-human, DABUS the AI-based machine, as inventor.
This case study currently represents the position of the law on this subject. AI systems are not recognized as authors under most legal systems, whether common law or civil law. Insofar as humans have contributed or have made the arrangements necessary for the creation of such works either by programming or writing software enabling AI to run, it is usually held that they qualify for authorship, not the machines.
We cannot stop technology from expanding and evolving. At best, we can only make our laws attune with it, in order to achieve perfect balance in society. With the current state of the law, it is not difficult to imagine that things may become more complicated in the future where automation of AI machines reaches a degree that no longer requires human input from start to finish.
The question therefore arises whether at such point, a different approach could be taken towards the protection of AI-generated creative works. For instance, it has been suggested that a sui generis class of rights be created for this purpose. Creating a new class of IP rights might be justified if it solves a practical problem in the stream of commerce. This, after all, is what the utilitarian economic justification of IP rights is based on. The existing framework could thus be slightly amended by granting a certain degree of exclusivity to AI-generated works. On a scale, society stands to benefit more from the new class of rights because such rights would create incentives to invest more in AI research and training, as well as share the results thereof with the public. In support of this position is former Director General of WIPO, Francis Gurry, who in an interview on the subject opined that “from a purely economic perspective, if we set aside other aims of the IP system such as ‘moral rights’, there is no reason why we shouldn’t use IP to reward AI-generated inventions or creations. But this still requires some thought.”,/p>
The second suggestion in this regard is for the removal of the requirement for an author or inventor to be identified during registration, thus shifting the focus towards the substantive aspects of the work itself. This revised registration process would instead prioritize the inclusion of accurate contact information and names or addresses for correspondence. A country that has adopted this method is South Africa. It was on this basis that the DABUS application, as elaborated upon above, was approved in South Africa after being filed cross-jurisdictionally; on the grounds that the patent office only checked for basic formal requirements.
The Intersection of AI and IP: Developments around the Globe
In the light of the foregoing issues, some nations around the world have begun the process of recognizing and regulating AI-generated works within their bounds. Some of them are briefly discussed below:
United States of America
In a guidance document released in 2023 titled “Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence”, the US Copyright Office reiterated the fact that works authored by non-humans, including AI-generated works, are not protectable under existing copyright law. The guide provided that the Office would evaluate AI-generated works by distinguishing human-authored works assisted by AI, from works where AI independently executes expressive elements. And that while the former would be granted copyright protection upon adherence with certain rules, the latter class with no sufficient human creative control over the AI’s expressive elements, are deemed unprotectable.
Nigeria has been actively involved in the field of AI, establishing the first Africa National Center for AI and Robotics and also gearing to roll out the National Policy on Artificial Intelligence. In the same vein, the impact of AI is also being felt in the nation’s creative industries. For instance, in January 2023, designer and filmmaker, Malik Afegbua, was able to come up with a fashion show made up of AI-generated images of older individuals.
Regardless of these developments and improvements, AI-generated works are not recognized as subjects of copyright or any other IP protection system in Nigeria.
South Africa has shown a favorable inclination towards AI-generated works in the sense that the nation’s Companies and Intellectual Property Commission (CIPC), which is in charge of its patent filings, is more concerned with the formalities of a work, the completion of the filing process and the novelty or technical application of the invention, over and above other requirements such as designation of author/inventor. It was on this basis that the DABUS application, as elaborated upon above, was approved in South Africa designating DABUS, the AI machine, as inventor.
Australia also presents another deviation from the norm. In 2021, its Federal High Court overruled the earlier decision of the Patents Commissioner, which had held that an AI machine did not qualify for patent. Although the judgment does not expressly grant a patent to DABUS (which was the AI inventor, in that case), it has set a precedent within the country for AI systems to be recognized as inventors while a human is listed as the assignee. In addition, the Australian government has also launched industry-wide consultation via a discussion paper titled “Safe and Responsible AI in Australia”, to address the potential risks of AI including IP infringement, and to establish responsible practices.
The need to find balance in all human endeavours is one that cannot be overstated. In this regard, there is need to strike a balance between the encouragement of innovation from all angles, whether human or machines, and the safeguarding of IP rights. This need to find balance becomes more imminent, especially as AI continues to progress, and automation continues to reach degrees which extend past the concept of human inventorship. This would occur for example, where computers can now deviate from the algorithm provided by a human, or relate inputs and outputs without instructions from a human.
Thus, there is need for ongoing discussions, legal/legislative adjustments, and policy interventions in order to achieve balance and address the challenges occasioned by the intersection of AI and IP. Such solutions would herald great benefits, encourage innovation from all sides and still safeguard IP rights and society’s welfare.
For further information on this article and area of law, Please contact Blessing Nwankwo at:
P. A. Ajibade & Co., Lagos by
Telephone (+234 1 472 9890), Fax (+234 1 4605092)