THE RAGHAV ISSUE: SHOULD AI BE GRANTED AUTHORSHIP RIGHTS? – American University Intellectual Property Brief

2022-03-25 10:11:36 By : Mr. HOLLIS HOU

Enter your email address to subscribe to this blog and receive notifications of new posts by email!

Recently, the Indian Copyright Office recognized AI as a co-author for a painting. However, the said authorities have sent a withdrawal notice to the human co-author. This article delineates the current legal status of AI’s authorship. It further analyzes the fundamental issues that must be resolved before granting such legal status and concludes by suggesting some plausible solutions to this AI-copyright conundrum.

Across the globe, the debate concerning legal recognition of Artificial Intelligence (AI) as a creator of work has developed remarkably, with the advancement in AI from simple machines to complicated human robots like Sophia. Recently, the Indian Copyright Office granted an AI, RAGHAV (Robust Artificial Intelligent Graphics and Art Visualizer), the status of co-author for a painting titled Suryast. However, after granting RAGHAV co-ownership, the Indian Copyright Office sent a notice of withdrawal to Mr. Ankit Sahni, the copyright holder and the other author of this artwork, asking him to inform the Copyright Office about the legal status is RAGHAV. Accordingly, the debate over AI copyright ownership in India has become much more interesting and the time is ripe to develop a solution.

Copyright authorship in India is governed by Section 2(d)(iii) and Section 2(d)(vi) of the 1957 Copyright Act. In the absence of explicit and precise legislation and jurisprudence regarding the ability to recognize an AI program as a co-author in the copyright law, this move by the copyright office was a bit shocking.

In light of this development, this article delineates AI’s legal status in India and around the globe to date. It further analyses the fundamental issues that must be resolved before granting AI such legal status. The article concludes by suggesting some plausible solutions to this AI-copyright conundrum.

Status Quo of AI Authorship: India & Beyond

In the India scenario, RAGHAV is the first AI entity to get co-author status. RAGHAV used the ‘Starry Night’ painting by Vincent van Gogh and a photograph clicked by Mr. Sahni to create the work of art at issue. This copyright registration was a classified submission by Mr. Sahni to the Indian Parliamentary Standing Committee on “protecting AI-created work as well as AI itself.”

On a bare perusal of the Copyright Act, it can be understood from section 2(d)(iii) that anybody who is an artist will be considered as an author under this provision. However, Section 2(d)(vi) further clarifies that the individual who causes an artistic work to be created by a computer shall be the author of that work. Through the literal interpretation of this provision, it can be implied that for an AI work (a computer-generated work), the one causing the work to be created, i.e. the creator of the AI, will be the author of that work. This interpretation is supported by the copyright office as it only allows natural or juristic persons to be authors.

The law seems prima facie settled with regards to this issue. However, the recent parliamentary report on Artificial Intelligence and Intellectual Property Rights (IPR) suggests that because AI and its applications in India will generate revenue, separate rights (para 8.2) should be given to AI entities. In its para 8.3, this report states that the existing Copyright Act is not well equipped to facilitate authorship and ownership by artificial intelligence; therefore, there is an urgent need to revisit this. The parliamentary committee report and the current legal framework deviates in viewpoint, thereby blurring India’s stance on whether to recognize AI as an author.

In the US, Chapter 202.02(b) of Compendium II of Copyright Office Practices warrants that the author of the work should be human. Similarly, in Acohs Pty Ltd v. Ucorp Pty Ltd, the Federal Court of Australia rejected copyright protection to a computer-generated work, noting human beings did not produce it.

Some countries have taken a different approach. In a few jurisdictions, the owner of the AI program is recognized as the author of the specific work. For instance, in Shenzhen Tencent v Shanghai Yingxun, China gave copyright protection to the creator of the AI tool for the literary works created by it. Other jurisdictions have actually acknowledged AI creators. For example, the Artificial Intelligence Virtual Artist (AIVA), an AI composing music, was recognized as a composer by the French and Luxembourg Author’s Rights Society. Similarly, the Device for the Autonomous Bootstrapping of Unified Sentience, better known as DABUS, was given the status of an inventor by the South African IP office after being rejected in the European Union and Australia IP offices. Also recently, Raghav was recognized as a co-author in Canada.

At the outset, we need to appreciate that AI and other technologies are becoming more capable and autonomous. As of now, there are three types of AI —Artificial Narrow Intelligence (ANI), Artificial General Intelligence (AGI), and Artificial Super Intelligence (ASI). ANI relies on humans for performing tasks, AGI is assumed to be at par with humans in terms of behaviors and intellect, and ASI is the hypothetical AI that surpasses the capacity of human intelligence and ability.

Globally, scientists have been unable to obtain any significant breakthrough in the spectrum of AGI and ASI, and presently we have ANIs only. Until and unless we have AI entities with emotions and intellect (like Rajni Kant’s Chitti from Robot), giving IP rights to them serves no purpose apart from acknowledgement because such AI tools cannot accrue any financial or emotional sense of satisfaction or benefits out of it. Therefore, it is imperative to bring into light some fundamental issues with the premise of granting authorship to AI entities.

If an AI entity is allotted the status of an author and a third party infringes the copyright, or AI infringes the copyright of a third party, then AI cannot sue or be sued. AI cannot be subject to legal proceedings as this status is only attributable to humans by virtue of their human nature; and unlike in a corporate structure, AI does not have a controlling human medium that could be subject to sanctions.

The incapability of AI to sue or be sued negates the rationale of copyright protection to safeguard original work from unauthorized uses. A work that cannot be sued for infringement will have the same effect as being in the public domain so granting such work the status of an author serves no purpose. Copyright law provides for both civil and criminal remedies. One cannot ask for civil remedies such as injunction, compensation, and damages from AI because the AI is just a machine without any assets or ability to comply with a judgment. AI is all operated with human collaboration, so in reality the criminal remedy must be sought after the human creator not the AI to have any impact. Further, humans may try to shift the blame on AI and absolve themselves of any criminal or civil liability, which is not at all desirable. This implies a lack of human control, and it is highly unwarranted. Therefore, the conundrum of IPR’s liability fetters its legal status as an author.

In most countries, the author of a work, being the first owner, is entitled to sue to enforce their rights, negotiate, license, and assign the copyright. The AI entity, by itself, is currently not in a position to perform any of these important functions; thus, rendering the ownership copyright provision redundant.

Certainly, the human behind the AI may authorize the necessary legal proceedings or transactions and hire a legal representative for the AI. However, one cannot plainly assume that AI has given consent to humans to make decisions on its behalf. AI itself cannot enforce its right, and the human is making decisions for the AI, so giving authorship rights to the AI will become a mere academic exercise and an impractical right to grant.

Perpetual Copyright Protection to AI-generated works

Copyright in India for any literary, dramatic, musical, or artistic work published within the author’s lifetime subsists for sixty years from the beginning of the calendar year following the year in which the author dies. If AI is granted such protection, then the copyrighted work shall never enter the public domain as the hardware of AI’s can be replaceable, and humans can update the software timely, meaning thereby, an AI will never die. Also, for joint authorship, the term is calculated from the time of the last author’s death. Therefore, such copyright protection will exist in perpetuity, thus totally defeating the purpose of enacting the law.

Further, though it can be argued that a simple amendment can resolve the issue of perpetuity,  such an amendment would require a comprehensive understanding and overhaul of the fundamentals of AI entities. Keeping in mind the ever longing debate of copyright protection, solving these issues will be an arduous task.

The main argument in favor of AI being recognized as an author is that denying it such status would result in its creations not qualifying as works and thereby precluding copyright protection for any such creations. Therefore, it will defeat the purpose of copyright law as nobody in the absence of protection will develop an AI capable of creation. The parliamentary committee’s report also argues on similar lines; however, it fails to explain how revisiting the law will generate revenue, incentivize and encourage creativity, and result in the development of more AI solutions.

Critics fail to understand how granting AI authorship will impede creativity. If an AI program is not recognized as an author of work but the human creators are given copyright ownership as authors, the human AI creator will be incentivized to improve the AI to further promote the progress of the sciences and useful arts, which fulfils the ultimate purpose of copyright law. However, if the AI is given authorship rights with no way to enforce those rights, copyright law provisions will be weakened, and creation of new works by AI will be discouraged. Thus, in the absence of an adequate justification by the parliamentary committee report, no action should be taken to change the current copyright laws to grant AI copyright authorship rights.

The authors suggest that AI’s work should be recognized to have contributed to the creation of work. The copyright office should continue to recognize only the human behind the creation as the author while issuing copyright participation certificates in favor of autonomous tools (like RAGHAV) to give it credit for its participation. As merely giving credit as the author does not amount to an acknowledgement of the copyright, this should be accepted.

Additionally, such inclusion will ensure that necessary credit is given to AI entities, their work is recognized for copyright protection, and the cumbersome process to change the laws in totality is avoided. Further, for any potential “bad act” of the AI, the controller of the AI should be liable under the principal-agent relationship with humans as principal and AI as an agent. Likewise, all the rights of the AI should be vested with the human under the principle of accession—the owner of the property should own all those things produced in his property.

If India wishes to revisit the law and change the existing jurisprudence to grant recognition to AI, then it will have to consider issues arising from all the possible AIs, including ASI, because changing the law again and again for each and every development is not a plausible solution. Therefore, rather than changing the copyright law now, the legislature should resolve the issues with the other possible existing solutions.