EXAMINING THE IMPLICATIONS OF AI-GENERATED CONTENTS ON COPYRIGGHT INFRINGEMENT IN NIGERIA

EXAMINING THE IMPLICATIONS OF AI-GENERATED CONTENTS ON COPYRIGGHT INFRINGEMENT IN NIGERIA

~ METI MONDAY UKPEH, ESQ.

Past Chairman, NBA-YLF, Uyo Branch 

Member, National NBA-YLF Pro Bono Committee 

Preceptor, Legal Digest.

Abstract

 Technological advancements and access to computing power has made it possible for machines to artificially perceive, think, understand, learn, produce and interact without explicit human programming. This new reality known as artificial intelligence was introduced by the English mathematician Alan Turing in 1950 and later coined by an American computer scientist John McCarthy during the Dartmouth Conference in 1956. This new reality has brought with it various legal and philosophical complexities bordering on copyright ownership. These challenges stem from the fact that extant laws and legal systems on copyright were built on the presumption that there is some form of human intervention. It becomes imperative to solve the problem of the ownership of copyright content created by machines. This includes establishing a legal framework for the purposes of some form of regulatory intervention in respect of regulating Artificial Intelligence. It is further suggested that where there is an artificial intelligence infringement on copyright, a human-based “smoking gun” is a prerequisite for liability and appropriate remedy. The approach adopted is analytical and comparative.

 

KEY WORDS: artificial intelligence, copyright, ownership, infringement, liabilities, etc.

Introduction

The past half a decade has seen a tremendous growth in the adoption of artificial intelligence technologies by average everyday individuals. This adoption has been fueled mainly by the meteoric advancement in the technical capabilities of AI systems within this time period. From text-based models like ChatGPT and Grok generating full length papers to image generators like Midjourney creating realistic imagery and art, AI has been deployed to a wide variety of use cases. These use cases, however, pose novel legal questions vis-à-vis existing IP laws globally. In this essay, with a focus on Nigerian intellectual property regulations, three key challenges will be examined: (a) whether the training of AI systems on copyrighted content constitutes copyright infringement, (b) who ownership in AI generated content vests and (c) who bears liability for infringing AI generated content.

Conceptual Framework:

Generative AI/ Ai-generated content is a program that creates works/ Ai-generated content in accordance with prompts from a user. Pictures, drawings, literary works, and even virtual worlds can be "generated" from this software. This low-effort tool has now become a facility through which anyone can be creative.

It is trained on large sets of data, and taught to identify patterns among them, and to create new and original data based on what it has learned- "new, unseen before samples that are not identical to but have the same characteristics as the input data1". It is, for example, trained on hundreds of books, so that it can, ideally, produce the same quality of books as pertains to diction, plot strength, and imagination.

"Copyright is a type of intellectual property that protects original works of authorship as soon as an author fixes the work in a tangible form of expression. In copyright law, there are different types of works, including paintings, photographs, illustrations, musical compositions, sound recordings, computer programs, books, poems, blog posts, movies, architectural works, plays, and so much more!"

The term "Copyright" refers to several rights that accrue to the author of an original literary work, that enables such author to control its use, distribution, and monetization among others. The duration of a copyright is the duration of the author's life plus 70 years after his death.

2.0. Training or Copyright Infringement?

AI systems, particularly the narrow generative AI systems that exist at present, are at a high level, easy enough to understand. They are trained on large sets of data using machine learning algorithms in order to “learn” how to generate seemingly intelligent output. To understand whether there is in fact infringement, it is necessary to first understand the process through which the typical generative AI system goes from training to prompt to output. Large Language Models(LLM) would be specifically discussed first since they deal mostly with literary content before other generative AI systems like image generators would be examined. Large Language Models(LLMs), as the name implies, are AI models trained on copious amounts of language data consisting articles, books, research papers and even social media posts to infer and understand the meaning and context of words. In generating responses to user prompts, these models generate words one at a time and work by predicting the next suitable word in word sequences. The question of whether infringement occurs at any point in this process is then raised -- if Meti Ukpeh writes a novel and that novel is then used as training data for an LLM, is there copyright infringement? 

Per the Copyright Act, copyright is infringed by any person who, inter alia, does or causes any person to do an act which constitutes a violation of any of the exclusive rights conferred under the Act. In the case of literary works, holders of copyright in such works are vested with the exclusive right to perform eleven(11) specific acts in relation to their work. However, of the eleven, only two are specifically relevant to the present discourse, viz: 

a. The right to reproduce, 

b. The right to create adaptations. 

a. The Right To Reproduce: 

Section 9(a) expressly provides that the exclusive right to reproduce a literary work shall vest in the copyright holder of such work. The term “reproduction” is defined in the Act as: 

the making of one or more copies of a literary, musical or artistic work, audiovisual work or sound recording.

 It therefore stands to reason that if a company, planning to use a copyrighted work to train its model, proceeds to make a copy of the said work at any point in order to effect such training, same would constitute infringement under section 9(a). However, where it is ensured that no unauthorized copy of such work is made, there would be no infringement since there is no reproduction stricto sensu. 

Another avenue through which infringement may occur is the AI system itself generating output that is a reproduction of its copyrighted training data. For instance, if an AI system is trained on copyrighted material discussing the history of the Second World War and it outputs content substantially reproducing the copyrighted work, this would constitute a reproduction under the Act and consequently, copyright infringement. However, the odds of an AI system substantially reproducing its copyrighted training data is close to zero. This is due to the fact that these generative AI systems function somewhat like the typical human brain and not at all like search engines as is often commonly misconceived. When prompts are inputted, the AI system generates brand new content based on the collective “knowledge” gained during the training phase from data that has long since been discarded. 

The process of training is a one off thing and whatever the AI system produces is often the product of wholly original artificial thought. Of course this position would be different in situations where the user expressly prompts the AI system to rely on specific material in producing content. For instance, in situations where an AI system is prompted to generate an abridged version of a copyrighted novel, there would be exist a strong argument for substantial reproduction. However, even in the exceptional circumstances where there appears to be substantial reproduction on the part of the AI system, the fair dealing exception may still likely apply to absolve liability. Fair dealing Fair dealing (otherwise referred to as fair use in some jurisdictions) is a defence to a copyright infringement action which permits individuals to use copyrighted material without the permission of the copyright holder for purposes such as: 

Private use 

Parody, satire, pastiche, or caricature 

Non-commercial research and private study

Criticism, review or the reporting of current events. 

It is necessary to clarify that this list is not exhaustive and acts which do not directly fall within any of the aforementioned purposes can still qualify as fair dealing. In assessing whether a specific use is fair, the Act outlines four factors to be taken into consideration: 

The purpose and character of its usage. 

The nature of the copyrighted work. 

The amount and substantiality of the portion used in relation to the copyrighted work as a whole. 

The effect of the use on the potential market or value of the copyrighted work. 

These factors will be analyzed seriatim within the broader context of AI training and content generation:

Purpose And Character Of Use: This factor primarily considers the purpose such work is utilized for (whether commercial or not) and the character of such usage (whether transformative or not). Where a work is used for a non-commercial purpose, it strengthens the argument for fair dealing. However, the mere fact that a work is utilized for a commercial purpose is not a dispositive answer to the question of whether or not such use is fair dealing. All four factors must be applied together. 

On the character of use question, where a work used in such a manner as to be considered transformative, it weighs heavily in favour of a finding for fair dealing. Therefore, where a person writes a review or critique of a work and in the process reproduces sections of the work in order to advance his arguments, such a use would be considered transformative regardless of the unauthorized reproduction. 

The Nature Of The Work: This factor examines the nature of the original work. Works considered highly creative are afforded better protection than those considered less creative. This reasoning follows common sense in that it would be unfair to punish an individual for simply reproducing factual information. If a person writes a non-fiction book discussing the history of the Second World War, they cannot go ahead to accuse all other books discussing the same subject as infringing on their work since there is no way such books could be written without putting forth similar facts to that contained in the original work. High creativity on the part of the original work weighs against a finding for fair dealing. 

Amount And Substantiality Of The Portion Used: The more of a work is reproduced, the less likely a court is to find fair dealing. The amount of a work copied is an important factor in determining whether a specific use is fair. Substantiality is also important in that even if a small portion of a work is reproduced but that portion happens to be a fundamental aspect of such a work, such use would weigh against a finding for fair dealing. 

Potential Market Effect and/or Value Of The Work: This factor takes into account the propensity of the use to adversely affect the market or value of the original work. If, for instance, a person publishes an abridged version of a book summarizing its most important points, such a work has the potential to negatively affect the market for the original work since people who buy the abridged version would be unlikely to still want the full original version. Where a use negatively affects the market for a work, it would be unlikely for it to be deemed fair dealing. 

Based on the preceding, when all four factors are taken together, it would appear that the question of whether or not an AI reproduction constitutes fair dealing would greatly depend on the facts and circumstances of each individual case. In circumstances where an AI system reproduces a small portion of a work in order to respond to an educational query, a good case for fair dealing can be made. Conversely, it would be quite difficult to justify an AI reproducing the lyrics of Ed Sheeran’s “Perfect” in its entirety to a user’s query requesting lyrics to create a song for commercial purposes. 

b. Right to Create Adaptations: Copyright holders possess, inter alia, the right to create adaptations of their work. Adaptation is defined by the Act as: 

the modification of a pre-existing work from one type of work to another or altering a work within the same type to make it suitable for different conditions of exploitation and may also involve altering the composition of the work.

 Two kinds of adaptation can be extracted from the above definition: 

 Modifying a work from one type of work to another (e.g literary work to artistic work and vice versa), 

Altering a work within the same type to make it suitable for different conditions of exploitation.

A close examination of the above definition would reveal that infringement under adaptation would only occur in very specific circumstances. The output commonly generated by AI systems gleaned from knowledge derived from the content of the copyrighted material it is trained on would be unlikely to be adjudged adaptations. If the converse were true, most materials in existence today would also constitute infringement in some form since all works fundamentally derive something from other works already in existence. Dealing with the specific circumstances where the output would be adjudged infringement.

Good examples falling within the first kind of adaptation - modifying from one type of work to another - would be situations where an AI generates a comic strip based off of a novel or creates an audiobook out of a written work on a user’s request. These examples fall within the first kind of adaptation because the work is modified from one type to an entirely new one. 

Examples within the second kind - altering a work within the same type - would be creating an unauthorized translation of a work or even generating a movie script based off of a novel. These examples fall within the second kind of adaptation because the finished derivative work is still of the same type as the original. AI Image Generators AI image generators behave similarly to LLMs in that they are both subsets of the broader generative AI family. While LLMs deal with generating text, image generators deal with generating artistic works from text prompts. Image generators are trained on millions of images paired with textual descriptions in order to learn what specific entities like animals, colors, etc., “look” like. Artists looking to enforce their copyright in artistic works suffer the same challenges as their literary counterparts. Unless the creator of the AI system makes an unauthorized copy of the artistic work in the process of training, an image produced by an AI system rarely copies enough of a work to constitute a reproduction neither does it fulfill the conditions necessary to come under the classification of an adaptation. However, in the off chance that AI generated content resembles a work closely enough to constitute a reproduction, it is still likely that the fair dealing exception would apply albeit depending greatly on the facts and circumstances of each individual case. 

The Author Problem in AI Generated Content So far, this article has referred to the possibility of an AI system infringing the copyright of other individuals. However, this is a legal impossibility as only persons, whether natural or juristic, can possess ownership of and infringe on copyright under Nigerian law. As AI systems are not recognized as legal entities under Nigerian law, the question then arises of who is liable for copyright infringement in AI generated content. 

Another pivotal question of who copyright in AI generated content vests is also raised. The answer to these questions not only affords the owners of such works due protection under Nigerian law but also clarifies the individual(s) to be held responsible should such works infringe on the copyright of others. On the question of authorship, section 28(1) of the Copyright Act provides that except as otherwise provided in an agreement, copyright conferred by the Act shall initially vest in the author. An author is clarified by the Act as being either a person possessing Nigerian citizenship or habitually resident in Nigeria or a body corporate incorporated under Nigerian law . It is therefore clear that generative AI systems are at present, unable to be conferred copyright protection as authors under current copyright laws as they are not juristic persons. As a result, the question of whether AI generated works can be afforded copyright protection therefore hinges on the question of whether or not the user who prompts such an AI system into generating a work can be considered the author of said work. 

Per Section 2(2) of the Act, copyright in literary, musical or artistic works shall not vest unless: 

Some effort has been expended in making the work, to give it an original character; and, 

The work has been fixated in a medium. 

Since the Act requires “some expended effort” in the creation of a work in order to make it eligible for copyright protection, it is only fair to assume that the person who expends that required effort would be considered the author of such a work. Following this logic, the earlier question of whether users who prompt AI systems can be considered authors of the generated content can then be reframed as whether the process of prompting a generative AI system fulfils the requirement of expending some effort sufficient to give the work an original character. In the course of answering this question of originality by various courts over time, two diverging doctrines have emerged: 

the “sweat of the brow” and 

“modicum of creativity” doctrines. 

Both will now be briefly examined in the context of generative AI systems. 

Sweat of the brow: This doctrine posits that a work is eligible for copyright protection so long as some level of skill and effort has been expended in its production. It is unnecessary that the work possess any element of creativity. The “sweat of the brow” doctrine originated in the UK in the early 20th century and is exemplified in the locus classicus, Walter v Lane , where the House of Lords held that works produced by the effort of some reporters in transcribing the verbal speeches of the Earl of Rosebery were eligible for copyright protection on the basis of the skill, effort and time expended in creating the work. 

Under this doctrine, therefore, works such as phone directories or compendia of facts and figures would be eligible for copyright on the basis of the skill, effort and time required in their production irrespective of the absence of creativity in them. This principle, in the context of prompting an AI system, can be argued both ways. While it is tempting to argue that the process of prompting should be considered sufficient to pass the effort criteria, this article adopts the contrary view. If a man asks another to paint a picture of a cat or write an article on a particular subject, who does authorship vest? The person who prompts or the person who creates? If in this scenario, authorship rightly vests in the person who creates, the answer need not be any different just because the creator happens to be a computer program. The effort expended via prompting does not directly go into the creation of the work but rather, guides the AI system into what the finished work should be. 

Modicum of creativity: This doctrine takes a different view to the traditional sweat of the brow doctrine and posits that a work must be independently created by the author and have a minimum level of creativity/intellectual judgment to be eligible for copyright protection. “Independent creation” simply means that the potential work must not copy any pre-existing work and must be brought to fruition by the independent efforts of the author. 

This principle was first propounded in the American case of Feist Publications, Inc. v Rural Telephone Service Co where the US Supreme Court held that telephone listings created by Rural and copied by Feist did not possess enough creativity to be afforded copyright protection. The modicum of creativity doctrine has become the standard for copyright eligibility in the United States as well as the EU. If this doctrine were to be applied to AI generated content, it would appear such works would fail to meet the independent creation requirement under the doctrine since AI generated works cannot be considered independently created by the user who prompts the AI system to generate them and would therefore, be ineligible for copyright protection. 

Bringing the discussion back to Nigeria, a close examination of the originality condition outlined in Section 2(2)(a) of the Copyright Act would favour the conclusion that the Act uses the sweat of the brow doctrine as opposed to the modicum of creativity doctrine. The section clearly states that a work would be ineligible for copyright protection unless “some effort has been expended in making the work, to give it an original character”. The clear emphasis on the effort expended in a work’s production is consistent with the reasoning behind the sweat of the brow doctrine. If this conclusion is to be followed, AI generated content would, nevertheless, still fail to enjoy copyright protection under Nigerian law flowing from the fact that AI systems are ineligible for copyright protection as authors under the Copyright Act. 

On the second question of who should be held liable for infringement: per the Copyright Act, infringement is primarily effected where a person “does or causes any person to do an act, which constitutes a violation of the exclusive rights conferred under this Act.” In cases where the infringement occurs in the training phase, there is no complexity as the AI’s creator would simply be held liable for the infringement. However, things become complex where the infringement occurs in the process of generating a response to a user’s query. If the AI system were simply taken as a tool, it can then be argued that users who wield the system as a tool to create infringing content would be liable in the same way a person who uses a photocopying machine as a tool to create unauthorized copies of a work would be. 

On the flip side, the counter argument can be made that as the creator of an AI system is the one who dictates what such a system is trained on and remains in control of the system, they should be held responsible for any infringement that occurs. This writer believes that it would be unfair to hold the users of AI systems liable for copyright infringement orchestrated by such systems by virtue of the fact that they are usually unaware and incapable of dictating how such AI systems function and operate. Furthermore, it is only the creators of such systems who can effect the changes necessary to ensure the copyright of existing works are not infringed by the content generated by such systems. 

The Data Privacy Problem: A preceding section of this article has shown that in the majority of use cases, AI systems do not, at least within the current IP regime, infringe on the copyright of individuals. However, the potential for data privacy breaches is still severe. Even in situations where AI systems do not breach copyright, they may still be in contravention of data privacy regulations where the personal data of individuals is in contention. This breach would usually occur in the training phase where content constituting personal data (such as social media profiles and posts for instance) is used in the AI’s training. To avoid this, companies training AI systems on personal data must ensure they possess adequate legal basis before processing such data.

While there are six(6) lawful bases for processing personal data available under Nigerian law (consent, performance of a contract to which the data subject is a party, protection of the vital interest of the data subject, compliance with a legal obligation, performance of a task carried out in public interest, and legitimate interest of the data controller), the safest basis is the consent of the data subject. Before personal data is used in training, the data subject’s consent for such training must be sought to ensure compliance with regulations currently in force. The data processing principles outlined in Section 24 of the Nigeria Data Protection Act (NDPA) must also be strictly adhered to. As long as these are done, it appears creators of AI systems would be safe in using content constituting personal data in an AI system’s training. 

3.0 Sample Cases of AI Infringement on Copyright:

In recent times, authors have instituted actions against Open AI, a private research laboratory that aims to develop AI to benefit humanity, and the creator of Chat GPT (generative AI that creates blocks of text like essays, ad copy, etc) claiming that not only were their works used to train Chat GPT, it was also plagiarizing their works while trying to produce its original works. Instances such as users trying to generate sequels and spinoffs to their favourite books have brought the conflict to the point of litigation by several famous authors.

The American Authors Guild and 17 other authors filed a lawsuit against Open AI for: copying their work wholesale, without permission or consideration, and feeding the copyrighted materials into large language models, which could lead to the creation of derivative works of fiction that are based on, mimics, summarizes or paraphrases their books. Additionally, these authors claimed that AI is being trained with pirated copies of their books.

Similar suits have sprung up throughout the year 2023, alleging similar claims against generative AI. There have also been reports that image-generating AI such as Midjourney, Stable Diffusion, and Dall-E are being used to generate images in the style of living artists, therefore, devaluing their human effort – the years it took the artists to formulate that style. These artists reason that this new development may displace them and their work due to how easy it is to replicate. More worrying is that US Copyright law does not sanction copying an artist's style, it only sanctions replicating his work.

Direct Copyright Infringement

In a direct infringement claim, the claimant argues their copyrighted work was used without permission to train AI models. This is governed by Section 36 (a) of the Copyright Act 2022. In 2023, artists Sarah Anderson, Kelly McKernan, and Karla Ortiz sued Stability AI, DeviantArt, Inc., and Midjourney, Inc., claiming their copyrighted images were used to train AI systems, resulting in infringing outputs

To succeed, claimants must prove valid copyright ownership and so, in determining which of Sarah Andersen’s works could be rightly claimed for in the case cited above, the court limited the works to only those that had been registered with the Copyright Office. Claimants must also demonstrate how their work was copied or used in training. Thus, the court ruled that Stability’s use of the plaintiffs’ copyrighted works to train its AI model, Stable Diffusion, amounted to direct infringement. However, claims against DeviantArt and Midjourney were dismissed, as the plaintiffs failed to show how the AI-generated images resembled the original works or what training, if any, occurred. The court allowed the plaintiffs to amend their claims to clarify facts showing how the developers had improperly used the plaintiffs’ copyrighted images.

Indire]ct Infringement

Generative AI is a self-evolving tool, which continues to learn and expand its training data set through user prompts, beyond the control of the developer who merely trains it to understand “how to learn”, a volitional requirement is unlikely to be met especially when the developer has taken all objectively plausible, practically feasible, and technologically available steps to ensure that the model does not bootleg output that is substantially similar to any of its input training data.

When direct infringement claims fail due to lack of causation/volition, the focus often shifts to the indirect infringement standard. Section 36 (e) of the Copyright Act provides that permitting for profit, any place for communications that infringe copyright would constitute indirect infringement, unless the operator was unaware or had no reasonable ground for believing such communication to be infringing.

Copyright Infringement through Derivative Works

Derivative works are covered by section 9 of the Copyright Act 2022 which are exclusive to the owner of such rights subject to the exceptions under Part II of the Act. Under the Nigerian Copyright Act, copyright holders have exclusive rights to reproduce, adapt, and distribute their works, and any unauthorized reproduction or adaptation constitutes infringement.

Creators also allege copyright infringement when AI-generated outputs produce what the creators deem to be derivative works of their copyrighted content. In their nature, GenAI models require large datasets for training, and when prompted, they may reproduce parts or entire works of copyrighted material. In cases involving derivative works, courts typically require claimants to prove that the AI outputs are direct copies or substantially similar to their copyrighted works. Failure to establish this often results in dismissal of the claims.

In 2023, authors Richard Kadrey, Sarah Silverman, and Christopher Golden filed a class action against Meta, accusing the company of using their copyrighted books to train its AI model, LLaMA, and claiming copyright infringement in the model’s outputs. The plaintiffs asserted direct and vicarious infringement, unfair competition, negligence, and unjust enrichment. However, the court dismissed most claims, including those related to the outputs, ruling that the plaintiffs failed to prove that the outputs were actual copies or substantially similar to their works to qualify as derivative works.

In early 2024, the Guangzhou Internet Court in China ruled in favour of a character licensing agency, affirming the rights of creators in a copyright infringement case involving AI. The claimant, holding the exclusive license for the Ultraman series in China, argued that an AI service provider’s model generated images identical to Ultraman works based on prompts related to “Ultraman.” The court upheld the claimant’s exclusive rights, ruling that the AI-generated images had some adaptation of the original work which constituted unauthorized derivative work.

If the output generated by the Generative AI model is materially similar or a mild alteration, of a work, it may infringe the rights of the copyright owner. However, who will be liable? The user or the Model developer? both? or neither? These questions are essential to consider given the Model itself lacks legal personhood for imputing any intent, liability, or damages – and responsibility for infringement has to be attributed to a human/corporation. The User’s interaction with the model by means of prompts outside the control of the developer is also a crucial factor to be considered when imputing liability.

The Defence of Fair Use:

In defending against copyright infringement claims, AI developers often cite fair use, which allows the unlicensed use of copyrighted works under certain conditions to promote freedom of expression. In Nigeria, the term is cited as “fair dealing,” and is permitted for purposes like private use, parody, satire, or caricature. The Nigerian Copyright Act outlines conditions for fair dealing, including: (i) purpose and character of the use, (ii) nature of the work, (iii) amount and substantiality used, and (iv) effect on the market or value of the work. These conditions are also part of U.S. Copyright Law, where an essential element is the transformative use of the work, meaning it is used in a new way or for a different purpose, and thus does not infringe its holder’s copyright.

In Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith, which involved Warhol’s Prince Series based on a photograph of the musician Prince by Lynn Goldsmith, Warhol argued that his alterations to the photograph through his paintings were transformative. However, the court held that minor alterations to a copyrighted work are not transformative under fair use where altered work was used commercially for substantially similar purpose as original.

Also, in June 2024, the Recording Industry Association of America (RIAA), on behalf of Sony Music Entertainment, Universal Music Group’s UMG Recordings, and Warner Records Inc., sued Suno and Udio, two AI models that produce AI-generated music. The plaintiffs alleged mass infringement of copyrighted sound recordings copied and exploited without permission to train the generative AI models. Admitting to using copyrighted songs to train their AI models, the defendants claimed that it was nothing more than a “fair” use of tools already available on the market. As the case progresses, it remains to be seen how the courts will interpret the law in reaching a decision.

4.0. Copyright of AI in selected Jurisdictions:

 The ownership of copyright works of artificial intelligence has been scrutinised in different jurisdictions. Examining these extant laws in these jurisdictions elucidates substantive rules that are applied to determine the adequacy of the legal frameworks. 

4.1 The United Kingdom: There were lots of legislative initiatives and proposals by the European Union (EU) and UK countries in 2017 with the aim of considering and addressing the impact of artificial intelligence on the society.These initiatives covered questions arising from liability, legal personality and other ethical and legal issues, including in the context of data processing. In March 2017, the UK Information Commissioner's Office updated its big data guidance to address the development of artificial intelligence and machine learning, and to provide GDPR, which will apply from 25 May 2018.In the UK, traditional copyright law grants protection to the original creations of authors (which include artists, composers and other creators). An author of a work is defined as the person who creates it; with additional clarification for particular types of work e.g. the producer of a sound recording is deemed to be its author. For a literary, artistic, dramatic or musical work, which includes software, to qualify for copyright protection the work must be "original". Case law provides that for a work to be original it must be its "author's own intellectual creation". The legal ownership of computer-generated works is perhaps deceptively straightforward in the UK. 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. The Act further defines a computer-generated work as one that “is generated by computer in circumstances such that there is no human author of the work.” This is the most concise definition given to computer generated works. This definition puts to rest any form of argument about creative works produced by artificial intelligent or any of its agents. This shows that the UK is among a few countries that offer a robust protection to computer generated works; the countries mentioned on the footnote also patterned theirs after the UK system of protection granted to computer generated works. This is one of the most salient aspects where UK and Irish copyright law diverges from the European norms.

4.2 United States: Under the US law, inventorship is the first point of analysis for determining ownership of IP. Identifying what contributed to the development of an AI-related patent for the purposes of determining whether someone was an "inventor" will probably happen more frequently. Although drawing the inventorship line may be complicated, the legal analysis substantially follows the legal touch points currently applied to other complex technologies. As AI develops, however, the patent bar may be confronted with another type of inventorship analysis that may be outside of the scope of current US law. Currently, inventors are individuals. But what if an AI-enabled machine invents something? What if an AI algorithm—without any human intervention—develops a new drug, a method of recognising diseases in medical images, or a new blade shape for a turbine? Section 100(f) of the Patent Act, 35 U.S.C.A. § 100(f) defines "inventor." Accordingly, perhaps Congress, and not the courts, may have to make changes to existing patent law to address potentially patentable subject matter developed autonomously by AI. Under section 101 of the Patent Act, the subject matter of a patent claim must be directed to a "process, machine, manufacture or composition of matter." However, the US Supreme Court held in Diamond v Diehr, that claims directed to nothing more than an abstract idea, such as a mathematical algorithm, or to natural phenomena or a law of nature are not eligible for patent protection. The US Patent Act limits patentable subject matter to “new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof”.Alice Corporation Pty. Ltd. v CLS Bank International recently made it more challenging for applicants to obtain patents on software or “computer-implemented inventions”. The seminal Alice decision has been interpreted and applied by the Federal Circuit and various lower federal district courts to generally exclude patent claims directed to subject matter that could be performed through an “ordinary mental process”, “in the human mind” or by “a human using a pen and paper”, with the limited exception for claims that specifically provide ways to achieve technological improvements over the tasks previously performed by people. This aspect of Alice’s legal framework creates tension with AI patents because the goal of AI is often to replicate human activity. For example, in Purepredictive Inc. v H20.AI, Inc., the United States District Court for the Northern District of California held that the asserted claims of US Patent No. 8,880,446 covering AIdriven predictive analytics were “directed to a mental process and the abstract concept of using mathematical algorithms to perform predictive analytics”. After further finding that the patent’s claims “do not make a specific improvement on an existing computer-related technology”, the court invalidated the claims for being directed to patent-ineligible subject matter.

 4.3 Nigeria: 

The Copyright Act 2022 does not refer to Generative AI, however, a close analysis seems to indicate an emphasis that could have bearing on any AI-generated work-related matter, – human effort. This requirement is very similar to the position of the US Copyright Office. According to s2(2) of the Act:

(2) Notwithstanding the provision of subsection (1), literary, musical or artistic work shall not be eligible for copyright unless — (a) some effort has been expended on making the work, to give it an original character ; and (b) the work has been fixed in any medium of expression known or later to be developed, from which it can be perceived, reproduced or otherwise communicated either directly or with the aid of any machine or device.

As seen above, section 2(2)(a) provides that for copyright to exist in a work, the author of a literary, musical or artistic work must have expended some effort in its creation, to give it an original character. The requirement in Section 2 (2) (b) that the work be fixed in any medium known or later to be developed is a proviso that directly accommodates AI-generated works.

The Nigerian Courts have not had the opportunity to dissect any matters regarding generative AI, but it is foreseeable that the courts will, in determining any future cases along the line of generative AI, adopt international standards and best practices regarding this matter.

The good news is, for one's copyright to subsist, the work need only be in existence (or fixed). However, to establish ownership, true publisher, originality, place of creation, and priority of creation, it is advised that creatives register their copyright with the National Copyright Commission, as section 43 of the Copyright Act provides that in an action for infringement of copyright, it will be presumed that:

copyright subsists in the registered work;

that the author on the work is the real author;

that the publisher of the work is the real publisher;

that the work is original – if the author is dead; and

that it was published or produced at the place or on a date appearing on the work.

 In Nigeria not every work is eligible for copyright. Simply put, an ostensible work of copyright must come under the list enumerated under the Copyright Act. These include literary works; musical works; artistic works; cinematograph films; sound recordings; and broadcasts. With respect to literal, musical or artistic work, sufficient efforts must have been expended by the creator to give them an original character and they must be fixed in any definite medium of expression now known or later to be developed from which it can be perceived, reproduced or otherwise communicated either directly or with the aid of any machine or device. 

A copyright protects the author. Under the Copyrights Act, the first ownership of copyright is enjoyed by the author of the work, unless a contract of employment or apprenticeship with a publisher stipulates that it belongs to the employer. An in-depth analysis exposes an interaction between Artificial Intelligence in Nigeria and the Copyright Act. AI related applications will usually run on software. However, it is important to note that there are no provisions for the protection of software under the Copyright Act in Nigeria. Copyright protection arguably only extends to the original documented expression of the software. This original expression does not extend to the functionality of the software. It is only limited to the blueprint either in an audio, written or any other form permissible by the Copyright Act as being protectable thereunder. 

5.0. Conclusion

 Although copyright laws have been moving away from originality standards that reward skill, labour and effort, perhaps we can establish an exception to that trend when it comes to the fruits of sophisticated artificial intelligence. The alternative seems contrary to the justifications for protecting creative works in the first place. Having critically analysed various laws above on the protection of AI, neither national nor international law recognises AI as a subject of law, which means that AI cannot be held personally liable for the damage it causes. In view of the foregoing, a question naturally arises: who is responsible for the damage caused by the actions of AI? In the absence of direct legal regulation of AI, a resort can be made to the United Nations Convention on the Use of Electronic Communications in International Contracts, which states that a person (whether a natural person or a legal entity) on whose behalf a computer was programmed should ultimately be responsible for any message generated by the machine. Such an interpretation complies with a general rule that the principal of a tool is responsible for the results obtained by the use of that tool since the tool has no independent volition of its own. 

This article has discussed the novel legal challenges raised by the meteoric advancement in AI technology. On the question of whether AI systems infringe on the copyright of other creators, this article has submitted that within existing Nigerian copyright laws, in the majority of situations, there would not be copyright infringement. On the question of authorship, this article has submitted that AI systems are the authors of content they generate and as they are not juristic persons, their works cannot be afforded copyright protection under the Copyright Act. On the question of who should be held liable for infringement, this article has submitted that creators of AI systems should be held liable for infringement orchestrated by such systems.

In concluding, it would be stated that there is a great need for new legislations regulating the creation, training and use of AI systems to be enacted to ensure that creators and users of AI systems are not allowed to profit off the labour of existing copyright holders as appears to be the unfortunate case at present. AI, and technological innovation at large, is changing the world. Law has a crucial role to play in ensuring this change is towards a world that is fairer, better and more equitable for al

6.0. Recommendations:

It has become imperative to amend existing IP legal regimes in order to provide for computer-generated works and Al operation, both in Nigeria and internationally. The law needs to clearly identify the rights and liabilities, if any, attributable to the Al for its inventions. Alternatively, a computer or Al related IP legislation may be enacted to deal with the intricacies associated with authorship of Al and computer-generated works, among others.

Another important aspect to consider is the use of class action lawsuits by claimants in copyright infringement cases. Given that GenAI can replicate works from numerous creators, class actions offer an efficient way for these creators to collectively seek relief. However, questions arise as to the potential for success if Nigeria were to face similar cases, particularly regarding the challenges of filing class actions and the dearth of cases in this regard. Expanding our jurisprudence on class actions, especially in technology-related disputes, would be beneficial.

A potential solution to address copyright owners’ concerns could be requiring GenAI developers to obtain licenses from copyright holders before training their models. While this may seem like a straightforward remedy for copyright infringement claims, the amount of data used to train generative AI makes this approach challenging. For instance, recent models like GPT are trained on trillions of words sourced from diverse materials. Obtaining licenses for such vast amounts of content, along with the associated costs, could hinder innovation and discourage the development and use of AI models, potentially stalling progress in the field.

The European Union Act on AI takes a proactive stance on copyright ownership, recognizing both the innovation potential of generative AI and the challenges it poses to creators. The Act requires authorization from copyright holders for any use of protected content, unless specific exceptions apply, such as for text and data mining. It also mandates AI model providers to ensure compliance with copyright laws, including transparency about the data used in training models, and to publicly disclose detailed summaries of the training content.

©2025