By Irfan Rashid
INTRODUCTION
The intersection of Artificial Intelligence (“AI”) and intellectual Property law is a dynamic and intricate field. The issues of intellectual property rights ownership, protection, and infringement are becoming increasingly important as AI systems develop. The subject of Al has become pertinent in today’s world while discussing Intellectual Property Rights (“IPRs”) because IPRs created by AI are real and no longer limited to works of literature or art that we may have seen or read twenty or fifty years ago. For instance, in 2018, a portrait made by AI was sold for 432,500 USD in New York.
Whether in the realm of graphic, visual, or even music to more complex literature or even software, the highest possible human factors involved are no longer necessary to produce original content. However, as AI increasingly finds a place in creative fields, it raises tough questions regarding copyright law. Some notable cases have brought this into focus. In 2023, a group of artists sued Stability AI, a company best known for its technology to generate images from textual descriptions. The artists asserted that Stability AI’s AI-generated artwork infringed on their copyrights because the AI was trained on datasets containing their works. Cases like this establish new precedents in determining fair use with regard to AI.
In this backdrop, basic questions arise: who owns the copyright in such works? Can AI itself be considered the author, or does copyright law still require human authorship? This article delves into a comparative analysis of how various jurisdictions tackle this impasse on the copyrightability of AI-generated works.
AI AND COPYRIGHT LAW IN INDIA
AI is struggling to be granted copyright in India because of Section 2(d) of The Copyright Act, 1957, which defines the term “author” and creates an obstacle to doing so. This section defines an “author” as one who “causes the creation of any literary, dramatic, musical, or aesthetic work, whether computer-generated”. Because of its emphasis on human agency, this concept prevents AI entities from being acknowledged as owners of copyrights. A traditional view of authorship is reflected in the provision, which begs the question of the legal standing of AI-generated works under Indian copyright laws. Furthermore, in accordance with Section 17 of The Copyright Act, the original owner of copyright is the creator of the work. And, when a work is produced under a contractual arrangement for consideration at the employer's request, ownership shifts to the employer. While the AI is commonly deemed the “agent” and the human developer its “principal”, legal and ethical debates emerge over the issue of ownership. Without the concept of personhood, AI is not entitled to the status of an employee, and this raises questions over the notion of authorship claimed by AI. As AI is designed and directed by human programmers to carry out specific tasks, it is often viewed as a tool or agent rather than an employee. This further solidifies the developer's authority and copyright over the AI’s creations.
In the case of Eastern Book Company & Ors v. D.B. Modak & Anr., the Supreme Court of India concluded that it is necessary to show that a derivative work is more than merely a copy of the original for it to be granted copyright protection. In addition to labour and capital, it must include the author's unique skill. The court defined judgement as the application of resources beyond the simple investment of labour and capital, and skill as those that are significant and nontrivial. Moreover, in the case of Amarnath Sehgal v. Union of India, Justice Pradeep Nandrajog acknowledged authors’ moral rights as protected by Section 57 of the Copyright Act, 1957. He underlined that authors have moral rights that grant them the ability to conserve, safeguard, and cultivate their works. Justice Nandrajog emphasized that the special power and mystery of original genius bestowed upon creative beings gives rise to rights like paternity, integrity preservation, and retraction. The court’s focus on the individual while debating the moral rights of writers implies that artificial beings are not included in the definition of authorship. As a result, identifying the creator of works produced by AI raises difficult issues with Indian copyright rules.
In 2020, artist-lawyer Ankit Sahni developed an AI tool (‘RAGHAV’) and used it to produce an artwork known as ‘Suryast’. Sahni applied for copyright registration claiming RAGHAV to be the sole author, but this application was rejected by the Indian Copyright Office on the ground that a human only can be afforded the title of authorship. Subsequently, Sahni filed another application listing himself and RAGHAV as co-authors. Interestingly, this was accepted. But, later, the Copyright Office issued a notice to withdraw the registration. This withdrawal by the copyright office prompted Sahni to challenge the removal with due process; hence, this opened critical viewpoints on the implications of recognizing AI in copyright law in India.
COPYRIGHT LAWS IN RELATION TO AI IN SEVERAL JURISDICTIONS
United Kingdom
As per Section 178 of United Kingdom’s Copyright, Deigns and Patents Act 1988, computer-generated works are those “generated by computer in circumstances such that there is no human author of the work”, and Section 9(3) states that in case of AI-generated literary, dramatic, musical or artistic work, the author shall be deemed to be the person who undertakes the required arrangements for the development or creation of such work. Similar to the UK, India defines the author of computer-generated literary, dramatic, musical, or creative works as the person who initiates or causes the work to be created under Section 2(d) (vi) of the Copyright Act. Thus, it's not very clear if AI-generated works can be given copyright protection but it seems that, much like the UK, Indian jurisprudence allows the Al systems' creator to benefit from copyright protection.
Furthermore, the UK High Court delved into the issue of authorship in the noteworthy case of Nova Production Ltd v. Mazooma Game Ltd.. The intricacies of the case revolved around electronic pool games and the computer-generated artwork displayed in individual frames on the screen. After a thorough analysis, the court determined that the programmer responsible for creating the program's distinctive features and software would be deemed as the author. The UK’s stance on this matter has been hailed by some as comprehensive and is even being proposed as a feasible approach for other countries to adopt.
United States of America
Analyzing various sections of the Compendium of U.S. Copyright Office Practices, Third Edition, it can be understood that the U.S. Copyright Office maintains a strict requirement for human authorship in order to grant copyright protection to a work. For instance, Section 306 makes it clear that the U.S. Copyright Office will only grant registration to works that are the product of a human creator. This provision highlights the importance of human creativity, as copyright protection is reserved for unique and original ideas of an individual's own making. Additionally, Section 313.2 mentions various types of works that do not fulfil the criterion of human authorship and are therefore not eligible for copyright protection. These include natural creations, works attributed to animals or plants, and pieces supposedly made by divine or supernatural entities. Furthermore, any works produced solely by machines or mechanical means, without any involvement of human creativity, are also not eligible for copyright protection.
In the case of Naruto v. Slater (also known as the “Monkey Selfie case”), the matter involving Naruto, a macaque who lives on an Indonesian island, began when nature photographer David Slater carelessly left his camera unattended on a wildlife reserve. To everyone’s surprise, Naruto the macaque took several stunning selfies using Slater’s camera. However, instead of acknowledging Naruto’s artistic efforts, Slater and his co-defendants Wildlife Personalities, Ltd. and Blurb, Inc. decided to publish a book featuring Naruto’s photographs under their own names and claiming copyright ownership over the works. Eventually, the court concluded that Naruto—and, more broadly, animals other than humans—lack statutory standing to sue under the Copyright Act. Here, animals and Als can be compared to non-human authors. Neither of them can be regarded as authors for the purposes of protectable works due to their inability to have rights or obligations.
The U.S. Copyright Office made history when in September 2022, Zarya of the Dawn, a graphic novel, became the first AI-assisted work to receive U.S. copyright. However, after a few months, this copyright registration was taken away partially by the Copyright Office though on account of AI-generated images not being human in authorship - one of the tenets of copyright. Although the text and the arrangement of elements were protected, the visuals created by the AI itself were not. This stance was later reaffirmed and supported by several courts in US.
To put it simply, AI-generated content is left in a legal limbo because the United States only grants copyright protection to works containing substantial human input.
Japan
The right to copyright AI output is not specified by the Japanese Copyright Act. Article 2(1)(i) defines “work” as a creatively produced expression of thoughts or sentiments that falls within the literary, academic, artistic, or musical domain. This indicates that the author can only be a human since AI lacks thoughts and sentiments. On the other hand, if an AI created a work based on human input and addressed any of the aforementioned ideas, the work would be protected by copyright. Moreover, as per Article 12(1), even a compilation (except a compilation that constitutes a database) should, by reason of the selection or arrangement of its contents, constitute an intellectual creation to be protected as a work. Because AI-generated output lacks direct human involvement in the creative process, it cannot meet the requirements under present Japanese law to be considered copyrightable. In this view, AI itself is not the author of its creations. Generally speaking, the copyright, if one exists, should go to the person or entity who is directing or controlling the output of the AI since this is what contributes the human element necessary for the work.
In 2018, Japan revised its Copyright Act and introduced Article 30-4 which allows for use of copyrighted works in carrying out information analysis, that is, training an AI model, without notice or permission from copyright owners. This is much more liberal than in many Western countries. However, there are limits. As long as AI training involved copyrighted works, the exemption would not be there if the output of the AI was substantial imitation or reproduction of the work. That’s because, in such a scenario, it would also amount to copyright infringement. Apart from that, the law has a ‘non-enjoyment’ clause wherein copyrighted materials are not permitted for mere experience or exploitation of the creative elements presented in the original works. If that happens in AI outputs by taking advantages of those creative elements, then it would amount to copyright infringement.
The exceptional case of ‘Sophia’ in Saudi Arabia
In the year of 2017, Saudi Arabia granted citizenship to an Al humanoid robot, Sophia. It is now the world’s first robot to be awarded citizenship. In this case, Sophia will be the author and owner of the work if the content qualifies for copyright. Since she is a legal entity and has produced content that is protected by a copyright, she should be the author and owner of the work. But since Al lacks legal personality, which is required to be able to assert legal rights, no court has ever given Al ownership. Granting legal personhood to AI could ensure accountability and ownership, but there are challenges in implementing this in practice. For instance, if AI generates defamatory content in India, it may not fulfill the defamation criteria of ‘intention to harm’. This indeed raises concerns about potential consequences. Although granting legal rights to AI requires the ability to enter into contracts, there are still practical barriers to overcome. There are debates over why legal entities and corporations have rights, but AI does not. But considering the current scenario, we can conclude that AI lacks legal personhood, and also, since Moral rights are related to the feelings and emotions of the human author, moral rights are not meant for AI. The decision to give legal status to Sophia is a fascinating development as it is encouraging discussions on the evolution of rights for AI.
CONCLUSION AND THE WAY FORWARD
As jurisdictions grapple with the complexities of AI-generated content, several key takeaways emerge: human authorship is still in itself a factor of the eligibility of work under most jurisdictions’ copyright laws, especially in the U.S., UK, and parts of Asia; legislative updates need to streamline rights and responsibilities within this fast-paced AI world; creators using those tools will need to document the creative process very thoroughly and claim ownership or authorship of the resulting material in case of a dispute; and multinational corporations have to determine each legal framework applicable around the world for AI-generated works.
A balanced approach is therefore necessary. An approach that would incentivize human creativity, protect original works, and allow for acknowledgement of the role of AI in generating new content. It is fair to state that since the current global legal landscape remains fragmented, this will remain one of the hot topics in the discussions on copyright law for years to come.
List of References:
1. Cohn, G. (2018, October 25). AI art at Christie’s sells for $432,500. The New York Times. Retrieved September 17, 2024, from https://www.nytimes.com/2018/10/25/arts/design/ai-art-sold-christies.html
2. Wodecki, B. (2024, August 19). Artists copyright claims against AI image generators advance in court. https://aibusiness.com/responsible-ai/artists-copyright-claims-against-ai-image-generators-advance-in-court
3. Williams, S. (2024, February 26). AI and Artists’ IP: Unpacking Copyright Infringement Allegations in Andersen v. Stability AI Ltd. Center for Art Law. https://itsartlaw.org/2024/02/26/artificial-intelligence-and-artists-intellectual-property-unpacking-copyright-infringement-allegations-in-andersen-v-stability-ai-ltd/
4. The Copyright Act, 1957, (1957). https://www.copyright.gov.in/Documents/Copyrightrules1957.pdf
5. Eastern Book Company & Ors vs D.B. Modak & Anr (2008) 1 SCC 1
6. Amar Nath Sehgal vs Union of India & Anr 2005 (30) PTC 253 (Del)
7. Dubey, R. (2023, March 31). Rise of the machines: AI and Copyright. Bar and Bench - Indian Legal News. https://www.barandbench.com/columns/rise-of-the-machines-ai-and-copyright
8. Sarkar, S. (2021, December 13). Exclusive: Indian Copyright Office issues withdrawal notice to AI co-author. MIP. https://www.managingip.com/article/2a5d0jj2zjo7fajsjwwlc/exclusive-indian-copyright-office-issues-withdrawal-notice-to-ai-co-author
9. Copyright, Designs and Patents Act 1988, (1988). https://www.legislation.gov.uk/ukpga/1988/48/data.pdf
10. Nova Productions Ltd v Mazooma Games Ltd & Ors. [2007] BusLR 1032. https://www.casemine.com https://www.casemine.com/judgement/uk/5a8ff71260d03e7f57ea71d6
11. Bhagat, S. COPYRIGHT PROTECTION FOR COMPUTER AND AI. Retrieved October 8, 2024, from https://supremoamicus.org/wp-content/uploads/2020/03/A40.v16.pdf
12. U.S. Copyright Office, Compendium Of U.S. Copyright Office Practices (3d ed. 2021). https://www.copyright.gov/comp3/docs/compendium.pdf
13. NARUTO V. SLATER, (U.S. Court of Appeals for the Ninth Circuit April 23, 2018). https://cdn.ca9.uscourts.gov/datastore/opinions/2018/04/23/16-15469.pdf
14. Glover, E. (2024, September 18). AI-Generated Content and Copyright Law: What we know. Built In. https://builtin.com/artificial-intelligence/ai-copyright
15. Copyright Act (Act No. 48 of 1970). Japanese Law Translation. https://www.japaneselawtranslation.go.jp/en/laws/view/1980/en
16. Copyright Law of Japan | Copyright Research and Information Center CRIC. (n.d.). https://www.cric.or.jp/english/clj/cl2.html
17. Stone, Z. (2017, November 7). Everything You Need To Know About Sophia, The World’s First Robot Citizen. Forbes. https://www.forbes.com/sites/zarastone/2017/11/07/everything-you-need-to-know-about-sophia-the-worlds-first-robot-citizen/
18. Augustian, A. (2022-2023). Authorship of AI-Generated Works: An Analytical Study. Indian Journal of Law and Legal Research, IV(VI), 5–11