GENERATIVE AI AND COPYRIGHT: ISSUES ARISING

Authored By: Favour Vincent Udoh

University of Uyo, Nigeria

INTRODUCTION

The evolution of artificial intelligence has become rampant and is gradually ‘wiping out’ human intelligence. It is safe to say that Artificial intelligence has revolutionized the whole world. At its core, AI encompasses a variety of technologies designed to perform tasks that traditionally require human intelligence. These technologies include machine learning algorithms, natural language processing, and predictive analytics, all of which are designed to learn from data, recognize patterns, and make informed decisions. Consequently, owing to the vast amounts of information generated, AI chatbots are even known to “hallucinate” false information creating problems for users of the technology, by generating everything from false narratives, to inaccurate citations for legal cases, with dramatic practical consequences while potentially infringing on others copyright. With its ability to analyze vast amounts of data, detect patterns, and make informed decisions, AI is reshaping every aspect of life. But the striking question posed is, ‘what happens when these technologies infringe on another’s intellectual property?’.

Whether Generative AI infringes an author’s copyright is a crucial issue that has to be settled as it is seemingly becoming recurrent in the globe today. The question as to who owns the copyright of a creative work when there is an involvement of Artificial Intelligence raises concerns and begs for answers. Thus, this paper seeks to conceptually analyse Artificial intelligence and copyright while addressing Copyright issues arising, and laws regulating them. It also seeks to offer useful recommendations and the way forward in the intersection of Generative AI and Copyright.

CONCEPTUAL CLARIFICATIONS

Artificial intelligence

Artificial Intelligence is the practice of getting machines to mimic human intelligence to perform tasks.^[1]In this order, machines are able to perform human tasks.

Generative AI

Generative AI also known as ‘Gen AI’ refers to deep-learning models that can generate high -quality text, images, and other content, based on the data they were trained on.^[2]It is powered by its latest large language model, can write poems, tell jokes, and can perform human tasks based on prompt (input).

What is Copyright?

A copyright is a collection of rights that automatically vests on someone who creates an original work of authorship such as literary work, story, movie or software. These rights include the right to reproduce the work, to prepare derivative work, distribute copies and to perform and display the work publicly.

The laws which protect Copyright are called ‘Copyright Laws ‘and they have been enacted into different states statute and legislations of a country’s jurisdiction. For instance, in Nigeria, the Copyright Act^[3] is the statute governing copyright production. The purpose of the aforementioned Act is to:

Protect the rights of authors and producers .
To ensure reward and recognition for authors’ intellectual property.
To provide appropriate limitations and exceptions to guarantee access to creative materials.

INTERSECTION BETWEEN GENERATIVE AI AND COPYRIGHT: ISSUES ARISING

The introduction of these technological machines and chatbots has brought to limelight, the need to clearly settle the concerns in relation to copyright, through legal instruments. As far as generative AI tools generate information from processed data, we cannot deny the chances of reproducing one’s intellectual property unduly, by the AI tools. Advertently, the growth of Generative AI has presented interesting challenges for copyright law globally and locally. The development of these generative AI tools brings higher proclivity for copyright infringement. Further, the legal person good of AI machines remains a fiction, and in the absence of legal certainty in that light, the question of copyrightability arises. Accordingly, two (2) issues which have come into consideration include:

The infringement of copyright by Generative AI tools;
Copyrightability of content produced using Generative AI.

We shall now address these issues in seriatim.

The Infringement of copyright by Generative AI tools

The issue of Generative AI infringing on the copyright of owners thrived when OpenAI a renowned and recognised tech company in collaboration with Microsoft introduced ChatGPT to the public.^[4] The New York Times popularly known as the NYT filed a federal lawsuit against OpenAI and Microsoft on Wednesday, Dec. 27, 2023, seeking to end the practice of using published material to train chatbots.^[5] The New York Times (NYT) core argument was that the data that powers ChatGPT has included millions of copyrighted works from the news organizations, articles that were used without consent or payment.^[6] By this singular act, people are not interested in purchasing magazines and products because there is an alternative ‘Artificial Intelligence’ , thus leading to a decrease and drastic drop in their market sales. OpenAI, argued that the vast amount of data used to train its artificial intelligence chatbot has been protected by “fair use” rules. The use of the term “fair use” has been carefully analysed by the American Courts.^[7]

The doctrine of Fair Use is a doctrine in American law that allows copyrighted material to be used for educational, research or commentary purposes. In the recent case of Andy Warhol Foundation for Visual Art Inc v. Goldsmith ^[8]professional photographer Lynn Goldsmith took copyrighted photographs of up-and-coming musician Prince Rogers Nelson. Goldsmith licensed a photograph to Vanity Fair to be used one time to create an illustration for one specific magazine issue. Artist Andy Warhol used Goldsmith’s photograph to create that licensed illustration and to make 15 additional, unlicensed illustrations, collectively called the Prince Series. Many years later, celebrity musician Prince died. By then, the Andy Warhol Foundation for the Visual Arts owned any copyrights in the Prince Series. Magazine publisher Condé Nast paid the foundation $10,000 to use a Prince Series illustration known as Orange Prince in a tribute issue. Goldsmith claimed that Orange Prince infringed on her copyright in the original photograph. The foundation argued that Orange Prince was a “fair”, “transformative” ^[9]use of Goldsmith’s photograph because the works convey different messages.

The Andy Warhol Foundation was found liable for Warhol’s unlicensed use of a photograph originally taken by Linda Goldsmith. It was opined that Warhol’s image was a transformative use, yet the Court disagreed and decided that this was not a fair use^[10].

It is worthy of note that the Fair use test implies that: the work in question has to have transformed the copyrighted work into something new, and the new work cannot compete with the original in the same marketplace, among other factors. The facts underlying the New York Times’s complaint centers around the use of copyrighted works in the development of generative AI tools, such as OpenAI’s ChatGPT and Microsoft’s Bing Chat (or “Copilot”), both of which are built on top of OpenAI’s GPT model. These tools are large language models (LLMs), which are built by “training” of the chat bots.

The models incorporate information from these datasets and “learn” the patterns of words within a given context. Then, when queried, the LLM(Large Language Model) can predict the most likely combination of words, generating a natural-language response to the user’s prompt.^[11]

The core allegation levelled against OpenAI by the New York Times is infringing on copyrights through the unlicensed and unauthorized use and reproduction of Times works during the training of its models. But OpenAI claims the works have been transformed into something new and peculiar, different from the Times publications. This is because:

LLMs sometimes “memorize” parts of the works included in training data. When this happens, the models can occasionally generate near-verbatim reproductions of the works.
LLMs produce “synthetic” search results that, when prompted, can reproduce “significantly more expressive content from [an] original article than what would traditionally be displayed” by an online search, effectively allowing readers to circumvent the Times’s paywall.

According to the New York Times, these problems present a significant threat to high-quality journalism. If readers can easily and at no cost generate summaries at their comfort zone and convenience or near-verbatim reproductions using GPT models, it will “obviate the need” to purchase access through the Times itself, thus affecting effective publications.^[12]

The context is not accessibility of summaries at one’s convenience but the copyrightability of the summaries generated by the Chatbots.

By doing so ,romantic authorship has been trampled upon because the originality and creativity of each work has been infringed by another.

This is the thin line that has been drawn.

It is pertinent to note that in recent times , NYT(New York Times) has not always displayed affection for romantic authorship as it has in this present case. In 1997, the Times found itself on the opposite side of a copyright suit^[13] when a group of freelance writers sued the publisher for including their articles in new digital archives without obtaining a proper permission. Thus, the sudden crave for romantic authorship creates doubts in suspicious minds including Open AI and Microsoft. In the case of New York Times v. Tasini^[14], the Times seemed not to value the “creative and deeply human” work of the authors that it relied as a justification for protecting its own copyrights in this current case New York Times v Microsoft ^[15]. Now it’s appears that reality has dawned on The New York Times when they sought for:

monetary damages ;and
a permanent injunction against further infringement.
Most importantly, the “destruction of all GPT or other Large Language Models and training sets that incorporate Times Works.”^[16]

Infringement of an author’s original work by chatbots amount to total disregard for ones Intellectual property. Although the Artificial Intelligence Act adopted by the European Union parliament is set to be fully swung in by 2026 but, till this time, there has been no comprehensive framework or exclusive policy applied in the development of AI tools.

Copyright laws varies between countries and these laws would clearly state when an infringement occurs and when it does not.

Copyrightability of Content produced using Generative AI:

Generative AI is capable of producing new types of “original” creations such as texts, images, and videos following prompts introduced manually by the user. Those creations occur as a result being trained with massive amounts of data, thus becoming destructive.

The issue here is “Whether works produced by Generative AI are copyright protected or whether copyright lies on the original author”?

AI generative software cannot create “original” images from zero, it needs a pre-existing data input to feed the algorithms through a machine learning model. The origin of those input is crucial not only regarding the ownership of the copyrights generated, but also to avoid potential right infringement of pre-existing works.^[17]

Internationally, for one to be afforded copyright protection, works and creations must be original. That is ,they must be the result of the author’s intellectualism and not a copy of a work in existence. These laws will vary based on the country’s jurisdiction.

In the Nigerian perspective, Section 2 of the Copyright Act^[18] states that:

(1) Subject to this Act, the following works shall be eligible for copyright :

(a) literary works ;

(b) musical works ;

(d) audiovisual works ;

(e) sound recordings ; and

(f ) broadcasts.

(2) Notwithstanding the provision of subsection (1), literary, musical or artistic work shall not be eligible for copyright unless —

(a) “some effort”^[19] 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.

(3) Any work that meets the requirements set out in subsection (2) shall be eligible for copyright, notwithstanding the quality of the work or the purpose for which the work was created.

From the foregoing, the word “some effort” could be interpreted from a human perspective because only humans expend effort. But the argument from writers is that, Artificial Intelligence also requires effort by the bots.

A careful perusal of Sect 108 of the Copyright Act^[20] will show whether Generative AI can be attributed as a person or not.

By virtue of Section 108,:

“4author” in the case of —

(a) audiovisual work means the person by whom the arrangements for the making of the audiovisual work were made, unless the parties to the making of the audiovisual work, provide otherwise by contract between themselves ;

(b) collective work, means the person responsible for the selection and arrangement of the collection ; Interpretation.

(d) sound recording, means the person by whom the arrangements for the making of the sound recording were made ; and

(e) a broadcast transmitted from within any country, means the person by whom the arrangements for the making or the transmission from within that country were undertaken.”

Sadly, the phrase “person” has not been adequately defined by the Act and this is a novel issue which we would hope that precedents would be set. The language of the copyright Act and the courts is that copyright should originate from the author, manifested by the expression of the author. In relation to Artificial Intelligence, the work is basically created by the computer program who enjoys neither legal nor corporate personality.

In Nigeria, the basis for the protection of copyright is not beyond works created by persons (natural or corporate) and there is no provision for works created by AI^[21]. This is evidenced by Section 5 of the Copyright Act, 2022 which states that:

“5. Copyright shall be conferred by this section on every work that is eligible, where the author or in the case of a work of joint authorship, any of the authors is at the time when the work or a substantial part made is —

(a) an individual who is a Nigerian citizen or is habitually resident in Nigeria ; or

(b) a body corporate incorporated by or under the laws of Nigeria.”

It has however been argued that the copyright of any work created by an AI program should be the work of the programmer as the copyright should belong to whoever has undertaken the necessary provisions for the creation of the work.

Therefore, copyright neither belongs to the robot or the artificial intelligence system but the person who created the robot or the intelligent system.^[22]

Buttressing this submission, the Nigerian Criminal Law describes a person to be held criminally liable as having a guilty mind(mens rea) and has done a guilty act(actus reus).

Apparently, AI does not have a mind of its own thus ,it cannot be held to be criminally responsible under Nigerian Criminal Law.

In summary, AI is not a person and does not fit in the interpretation of “person” under the Nigerian Copyright Therefore, copyrights cannot be vested in Artificial Intelligence but a human being.

RECOMMENDATION

Whether works created by Artificial Intelligence tools can be afforded copyright protection is a pressing issue in Nigeria and globally.

As such, certain recommendations are offered in this paper.

Comprehensive legislation should be enacted with adequate interpretation of specific phrases to set a clearer point:

Different countries should endeavor to enact clearer laws with less ambiguity, to ensure that necessary interpretation of keywords are explained in detail even without the help of the courts to avoid increasing lawsuits being filed daily.

Strict compliance and maintenance of policies by machine language models:

To avoid infringement of an author’s work, policies should be set out and a strict compliance of these policies would prevent copyright infringement. AI developers should ensure they are in compliance with the laws in regard to acquisition of data being used to train their models.

These policies would also serve as deterrence to others if they fail to comply with these policies.

AI policies on how data may be used, shared, transformed and inferred should be generated and strictly complied with.
The rights of an Author must be respected. Thus, Intellectual property owners should be fully compensated for their creation which has been transformed by chatbots into ‘new’ information.
Individuals, tech driven companies, Generative AI developers should stay abreast with the laws as it keeps evolving.

CONCLUSION

The past years has had a trail of technological developments, but generative AI is qualitatively different from everything that has come before.

Apparently, Copyright law is a sword that is going to hang over the heads of AI companies for several years unless they can bring out a tangible solution.

Technology is evolving at a very fast pace, and if not handled well, it will be dangerous. Therefore, there is an urgent need to provide a workable legal framework for AI locally and globally to avoid infringement of copyrighted works or creations.

Reference (S):

^[1]McKinsey and company, “What is Generative AI” published (2024).Available at https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai accessed 10th March,2025.

^[2] Kim Martineau, “What is Generative AI?” (2023). Available at https://research.ibm.com/blog/what-is-generative-AI accessed 10th March,2025.

^[4]Audrey Pope: “NYT v Open AI. The Times’s about face” in Harvard Law Review (2024). Available at https://harvardlawreview.org/blog/2024/04/nyt-v-openai-the-timess-about-face/#:~:text=Ultimately%2C%20the%20Supreme%20Court%20found,have%20infringed%20the%20freelancers’%20copyrights accessed 2nd March,2025.

^[5]Ibid

^[6] Ibid

^[7] Andy Warhol Foundation for Visual Art Inc v. Goldsmith (2023) 598 U.S. 508, 143 S. Ct. 1258, 215 L. Ed. 2d 473

^[8] Ibid Y

^[9] Ibid (emphasis added)

^[10] (emphasis added)

^[11] N.Y. Times Co. v. Microsoft Corp., No. 23-cv-11195, complaint filed, (S.D.N.Y. Dec. 27, 2023)

^[12]Audrey Pope: “NYT v Open AI. The Times’s about face” in Harvard Law Review (2024). Available at https://harvardlawreview.org/blog/2024/04/nyt-v-openai-the-timess-about-face/#:~:text=Ultimately%2C%20the%20Supreme%20Court%20found,have%20infringed%20the%20freelancers’%20copyrights accessed 10th March,2025.

^[13] New York Times Company Inc v Tasini(2003) 533 US 483

^[14] Ibid

^[15] N.Y. Times Co. v. Microsoft Corp., No. 23-cv-11195, complaint filed, (S.D.N.Y. Dec. 27, 2023)

^[16] Audrey Pope, “NYT v. OpenAI: The Times’s About-Face” Harvard Law Review(accessed 2nd March 2025).

^[17] European Innovation Council and SMEs Executive Agency, “Artificial Intelligence and Copyright: Use of Generative AI tools to develop new content” (2024). Available at https://intellectual-property-helpdesk.ec.europa.eu/news-events/news/artificial-intelligence-and-copyright-use-generative-ai-tools-develop-new-content-2024-07-16-0_en accessed 10th March,2025.

^[19] Ibid (emphasis added)

^[20] Ibid

^[21] Aminu Hassan, “The Legal, Ethical issues And impact of Artificial Intelligence on Legal Profession: Which Way Nigeria?” (accessed 2nd March 2025).

^[22] Ibid