Introduction
Over the last two years, the world has seen some groundbreaking developments in the field of Artificial Intelligence (‘AI’). Its growth has been unprecedented with AI companies coming up with new softwares at an exceptionally rapid pace in this “out-of-control race”. It has evolved from just being a tool at the hands of the creator to being one itself. Nations globally have been grappling with its impact, finding their laws inadequate when applied in this evolving context.
As AI continues to redefine creativity and production, questions arise about how traditional standards like originality should adapt to this new paradigm in the field of copyright law. Although the originality threshold has notably been lenient, granting entry to a multitude of works into the domain of copyright raises questions as to whether it can be potentially lowered to the extent of merely submitting an instruction, direction, or prompt for AI to generate the work.
Is the Indian Copyright Act well-equipped to protect and manage AI generated works? This paper tries to answer the same with a comprehensive discussion on the recent utilisation of AI in the domain of music composition and production and its implications on the copyright laws in the country.
Generative AI and the Music Industry
The music business is well acquainted with the complexities of copyright matters. Before it could fully recover from the setbacks caused by illegal file sharing, the industry has been faced by an even more disruptive technological phenomenon: the widespread emergence of AI. This new development has not only resulted in infringement concerns but have also raised questions about the ownership of the work. Recently, The Beatles experimented with AI to come up with their new song “Now and Then”. In this context, Josh Antonuccio, Associate Professor at Ohio University School of Media Arts and Studies stated:
“It is a tool set that is far beyond the scope of anything we have seen in human innovation to this point. The Beatles story is indicative of change for a lot of things in music, and that change will be across the board in creative industries.”
AI apps work with a database and algorithms fed to it by its creator. Based on those algorithms and data, it comes up with solutions to the problems faced by its users. It not only assists the users but also, in the process, creates new works using the knowledge it gathers from the data fed to it.
A song constitutes various elements like its lyrics, the music given to it by the composer, the voice of the singer and the recording produced by a producer. With AI, a person can do all these tasks with a single prompt. In essence, there are two distinct potential outcomes. One scenario involves the AI compiling various elements from different songs within its database to create a composition, raising significant copyright concerns due to its blatant replication. Conversely, the alternative outcome entails the AI generating a completely original piece, free from any direct replication. The problems faced by the first scenario are quite similar to the interplay between remix music produced by musicians and the copyright laws. However, the second scenario is a new development with the AI primarily performing the work of an author.
Copyright is a legal entitlement bestowed upon authors of literary, dramatic, musical, and artistic works, as well as producers of sound recordings. It comprises a range of rights, encompassing reproduction, public communication, adaptation, and translation of the work, among others which are provided to the authors. The laws of intellectual property in India do not recognise AI as an entity which can ask for protection.
Navigating the Copyright Act: Implications foe AI-Generated Content
Authorship in the Age of AI
Section 2(d) of the Indian Copyright Act includes in the definition of “author” as any person who causes the work to be generated by a computer, thus eliminating any chances of machines getting authorship of the work developed by it, independent of any human interference. This implies that none of the involved parties—the inventor of the machine, the machine itself or even, to some extent the copyright holders of the music in the machine’s database—would be eligible for protection. Instead, only the individual who simply provided a prompt to the AI to create the music would be entitled to any form of legal protection. Alternatively, even if we compare the relationship between the AI and the user to that of an employer and an employee for the ease of interpretation, this analogy loses relevance given that AI lacks the capacity to consent or authorise such relationships in any contractual context. This situation could also result in co-authorship claims by the developers of the AI considering their contribution in the “conception” phase of the work.
Another important factor to be considered while considering authorship is the change in the nature of help AI provides in case of AI-generated content. There is a difference between computer aided works and computer-generated works. The Indian jurisprudence affords protection only to the first category of works.
The Need for Originality and Creativity
For a musical work to qualify for protection in India, it has to be executed in the form of some kind of expression as copyright does not apply to ideas.
Furthermore, Section 13 of the Act states the “originality” of a work as a mandatory requirement to be eligible for copyright. The term “original work” lacks a precise definition. However, courts commonly examine the intrinsic connection between the idea and its expression in this respect, often referred to as the Doctrine of Merger. They consider whether the work was produced through the author’s skill and labour and the level of creativity present in the work. Courts distinguish between works created solely through labour and those that involve both skill and judgment.
AI lacks the conventional human components of creativity, intuition, and judgement and instead functions on algorithms and data processing. Even if AI can produce music that sounds like human compositions, its method of doing so does not fit with authorship and creative labour as often understood. The Indian courts’ interpretation of this originality has been in line with the courts in the US courts which follow the approach of ‘Modicum of creativity’. As per the doctrine, only those works fulfil the threshold of originality in which is a “product of exercise of skill and judgement’. In this context, the distinctive characteristics of AI innovation and the conventional legal system will be difficult for the courts to reconcile considering beyond its traditional role as a mere assistant, AI now stands as a creator of original works.
Ethical Dilemmas and Copyright Infringement
AI operates on the foundation of the large language model, functioning by processing the data provided to it by its creator using sophisticated algorithms. Traditionally, in case of a remix of a previously copyright protected song or music, the individual has to seek permission from the owner of the same to utilise it and make alterations, if any. They are subject to section 52(1)(j) of the Act. However, in case of AI generated music, the user has no permissions. The creator of the AI takes no such consent before feeding such data into the AI and the AI is incapable of doing so. This raises the question of who would bear the liability in case someone’s else’s work is utilized by virtue of being a part of the database available to the AI. It remains unclear as to who would bear liability in such a scenario if someone utilises another person’s work.
Identifying AI as a separate entity
Assigning AI a separate identity seems to be the answer to all such problems. However, assigning identity and ownership to AI presents challenges due to its lack of agency in contractual matters. Furthermore, moral rights granted to authors under Section 57 of the Act, such as the right to paternity and integrity, may prove difficult to enforce in case of AI-generated works, given AI’s inability to discern infringements affecting its reputation. Additionally, the mandatory requirement under the Act for the authors to accept royalty becomes complex, as AI lacks the capacity to claim or determine royalties under existing laws. Additionally, holding AI accountable for objectionable content poses significant challenges, as AI lacks the capacity for moral judgment. Thus, while recognizing AI as an author could introduce new dimensions to copyright law, numerous legal and practical hurdles must be addressed to ensure effective implementation.
Charting a Path Forward for Copyright Law
One of the potential options available to the legislation makers is to declare all works generated by AI which is openly accessible to everyone not eligible for any kind of protection under the copyright laws. Whether an AI-generated work should be protectable or not depends on the costs and the benefits attributable to providing protection. Works that are completely generated by AI should not give any of the users the right to prevent its production and distribution. The same should be subject to the decisions of the court on an ad-hoc basis to understand the contribution of the person claiming copyright in the process of producing the work. The lawmakers in India could also look beyond their country and consider what other nations have been doing. This problem is a global one and the EU and US doctrines like the Text and Data Mining (TDM) exceptions or fair use could provide some assistance in this regard. Text and Data Mining refers to the automated usage of data sets or information available to the user to analyse trends and produce useful results. This kind of usage qualifies as an exception to copyright infringement only in some legally defined situations. In some of these situations, the authors of those works can also choose to opt out of this system making the exception subject to their passive acceptance. The ‘fair use’ doctrine in the US permits an individual to use copyrighted material without asking for the author’s permission in that regard. Whether a work qualifies to be an exception under the same is decided based on a bunch of factors like the nature of copyrighted work, the purpose of the use, its effect on the copyrighted work and the substantiality of the portion used. These doctrines are not exclusive of each other. For instance, data utilised for scientific research has qualified to be an exception under both these doctrines where the author’s permission to utilise their data is not required to be taken. In EU, a separate act catering specifically to AI is also in its last stage. Similarly, India could consider introducing a sui generis system if IP rights for AI-generated content to adjust innovation incentives for AI. A special legislation specifically addressing the legal concerns which arise with the advent of AI is the way forward considering the redundancy of the current legislation of the country in answering these questions. The AI Act, pending approvals in the EU, is one of the first-ever legal framework regulating the use of AI in the Union.
Alternatively, the AI developers could also consider the method of audio steganography or citations in the work which has been “copied” from any of the works in its database. Audio steganography is a method of transmitting information or data with some hidden texts or audio information. They can be used to embed digital watermarks in the works of other artists. Additionally, laws or cases for that matter should also decide whether storing different works on the AI’s database amounts to fair use under the Act.
Conclusion
Fast advances in AI have catapulted us into unknown territory, converting it from a basic tool to a producer of unique creations. However, this progress has highlighted the shortcomings of current legal frameworks, notably in the area of copyright law. While the Indian Copyright Act offers some protection for AI-generated works, it fails to meet the technology’s particular difficulties. When AI gets engaged, the fundamental necessity of copyright protection is brought into doubt. Can a work developed exclusively through algorithms and data processing really be deemed original? The answer to this question remains unclear as courts deal with the intricacies of AI-generated content. By exploring these issues and proposing thoughtful solutions, the country can navigate the intersection of AI and copyright law.
This blog is written by Ankita Jagnani, 3rd Year Student, 5-year B.A. L.L.B.