I. The Rise of AI in Music Industry

Music is a deeply human form of expression, often considered as a mocktail of lived experiences, emotions, and creative intent. A song is an inseparable aspect of the person who composed, performed, or sang it, but increasing use of artificial intelligence in music production challenges this assumption. As algorithms begin to generate, replicate, and modify music, they challenge how creativity is defined and the long-standing ideas about artistic authorship.

Recent developments in India reflect how visible this revolution is. Mohit Suri’s Saiyaara generated widespread online engagement, particularly for its emotionally resonant soundtrack. While the original music album had already gone viral, a video recreating one of the songs in the voice of Kishore Kumar using artificial intelligence gained countrywide success and praise. This incident highlights how seamlessly artificial intelligence can be trained to imitate distinctive artistic voices, often without audiences clearly being able to differentiate between human and machine-created performance.

Artificial intelligence has undoubtedly pushed the boundaries of music creation further, lowering the entry barriers for aspiring creators but also at the same time fuelling the anxieties about the displacement of human creativity, commodification of artistic expression, and the unauthorised use of existing works for data training. It is now crystal clear that AI-generated music is no longer in its experimental phase; it is now commercially viable, widely consumed, and increasingly undistinguishable from human-made works, a reality evidenced by chart-performing AI tracks and industry-wide licensing disputes.  Its abilities are no longer restricted to assisting musicians, and it has moved ahead to independently generate complete music works. Technologies such as MuseNet, Jukebox, and MusicLM are capable of producing full compositions, simulating voices, and replicating musical styles with minimal human input.

As a result of such a technological evolution, certain legal problems have come to the forefront, exposing the inability of existing copyright frameworks to respond to creative output that no longer fits within the threshold set by human authorship models. It raises a major question, i.e., can copyright law, which assumes human authorship, effectively regulate AI-generated music without putting human artists and creative fairness at a disadvantage?

II. Questioning authorship

The question of authorship in AI-generated music sits in a legal no man’s land. With the increasing use of generative technologies, the distinction between human and machine creativity has blurred. Another example of this fascinating yet confusing intersection of AI in music is the viral release of Heart on My Sleeve in 2023, an AI-generated track that imitated the voices of Drake and The Weeknd. With authorship taking a centre seat in this intersection, the claims over it revolve around three possible actors: the user who provides prompts, the developer who designs the system, or, in the absence of either, no author at all.

This uncertainty is no longer theoretical. Authorship determines who can claim rights, control usage, receive royalties, and enforce protection under copyright law. So, when the authorship itself becomes indeterminate, the entire copyright structure is destabilized and complicates questions of accountability. This is highlighted by a surge in lawsuits brought by music companies against AI developers for unauthorized use of lyrics and sound recordings, reinforcing the inadequacy of authorship-based analysis in the current law.

India finds itself mired. The Indian Copyright Act, 1957,[1] is firmly grounded in the assumption of human creativity, upholding the ‘modicum of creativity’ doctrine and rejecting labour, diligence, and effort invested in a work as the sole criterion for copyright protection, as held by the Supreme Court in Eastern Book Company v. D.B. Modak (2008) [2]. Autonomous AI systems cannot consent to licensing, assign rights, or exercise moral claims. As a result, because of no clear statutory framework for issues like fair dealing in AI training, data mining, or the distinction between inspiration and infringement, the courts are grappling with this structural gap, as currently seen in the ANI v. OpenAI case [3], where music labels like T-Series and Saregama have intervened to challenge the unauthorized use of copyrighted songs for AI training, raising questions about infringement, fair use, and ownership.

India is not unique to this problem; jurisdictions across the globe are dealing with similar issues. In the United States, courts have consistently rejected protection for works lacking human authorship, as reaffirmed in Thaler v. Perlmutter (2023) [4]. Similarly, other jurisdictions have adopted a similar incoherent approach of either denying protection to purely machine-generated works or relying on legal fictions to assign authorship. This underscored a shared global difficulty: copyright law remains bound to human authorship, while creative production is not. The problem does not only lie in the ability to define ‘author’ but also in the inadequacy of the framework for regulating machine-generated activity despite a technological revolution.

III. Legal perspectives

The copyright law of India has been built on the assumption that creative work is a product of human intelligence, will and character. When AI-generated music challenges the traditional law, the problem no longer persists to authorship but also the conditions on which copyright protection is ensured. There is a need to consider the substantive needs of the copyright law with this emerging AI-generated music and the ways in which various jurisdiction have endeavoured, often inconsistent results, to apply them to AI-generated creativity.

At the core of copyright law, three conditions consistently contained in formulations are: (1) an element of originality that implies independent derivation of the work with some minimal level of creativity; (2) fixation, which ensures that the work is reduced to a tangible medium of expression; and (3) human authorship, as copyright law does not recognise works created solely by non-human entities without meaningful human creative contribution. Considering AI-generated music can be easily recorded and stored, fixation is not too challenging, but originality and authorship is particularly problematic. The AI-generated music strains these requirements and fall into a grey zone.

However, the copyright law of India under Section 2(d) of the 1957 Act[5] grants protection exclusively to works created by human authors, excluding AI-generated works unless they embody a sufficient degree of human creative input. This gap has led to the suggestion to revisit the scope of Section 13 and 17 of the Copyright Act, 1957, which address what works qualify to be given copyright and by whom the first authorship is vested.

This tension between human-authorship requirements and AI-generated output is not confined to India but extends across jurisdictions. The U.S. court in DC Comics v. Mark Towle (2015) [6] held that the Batmobile was a copyrightable character since it had a distinct and consistent creative identity created by human beings, although it was different in its appearance in various models. This shows that copyright law is only capable of protecting those works when they can be identified as products of human creativity. The Copyright Office of the United States confirmed in its January 2025 Report on AI in copyright that AI-generated material alone is not copyrightable, unless there has been a meaningful human contribution. The report stresses that thorough prompts in themselves, which are used to control the output, are not authorship unless a human rewrites or substantially edits the result of the AI output. This leaves a gap regarding the threshold of “meaningful” human involvement and the treatment of economically valuable AI-generated works that fall outside copyright protection.

In UK, the Copyright, Designs and Patents Act of 1988, Section 9(3) [7], gives the title of author of fully computer-generated works to the person who made the arrangements necessary to have the work created. Yet the UK courts tend firstly to seek a human author based on free and creative choices and personal touch, resorting to Section 9(3) only when no human author could be determined, as illustrated in Nova Productions v. Mazooma Games (2007), where the court emphasised human creative input as the primary basis for authorship [8]. This gap in regulatory coverage has enabled AI-generated music to proliferate without adequate legal oversight, reshaping the music industry and artistic labour. Therefore, it is mandatory to examine the economic as well as the cultural effect of the AI-generated music on artists and the industry.

IV. Accountability and fair use doctrine

Musical compositions created by AI pose new copyright-related issues, especially in cases that involve replicas of existing compositions or compositions trained using unlicensed data sets. American statute introduces the possibility of liability in the case of AI compositions that are substantially similar to existing protected works; the EU also takes into consideration the originality and fair use of datasets under fair dealing. At the centre of this argument is the question of whether artificial intelligence is simply an extrapolation of patterns or the unauthorised reproduction of creativity.

Sony Music and other big labels have threatened with legal action against AI companies that have allegedly scraped catalogues without their consent, treating such activities as mass piracy rather than innovation. Although fair use could be used in some minor transformative applications, there is no unity amongst courts in declaring the use of copyrighted works as AI training data transformative or exploitive. Such writs of passage also serve as reminders of why there is such an urgent need to design legal standards clarifying liability, defences, and safe harbour to balance innovation and the protection of artistic rights.

V. Artist opinions and ethical concerns

With the rise of AI-generated music, debates over creativity, authorship, and technology have left the music community divided: some embrace AI as a generative tool expanding human expression, while others fear it as a threat to artistic labour, eroding autonomy and commercialising uniquely human art to great extent.

This polarity is most visible in remix culture. Sampling and reinterpretation were long accepted as legitimate artistry; now AI automates them, producing mashups, reimagined vocals, and derivatives with unprecedented ease. But what was once curation has become calibration, guided by algorithms and blurring homage, theft, and innovation.

For many, art requires lived experience and emotion. Some listeners admire AI compositions, while others reject them as soulless. So, acceptance of such music depends on psychology and perception. Artists, like Hans Zimmer, have argued that while AI replicates sound, it cannot tell stories or the essence of music. Artists like Holly Herndon demonstrate a middle path—her album Proto (2019) used AI-trained voices as instruments while retaining human creative direction, distinguishing between sound production and emotional expression.

Acceptance ultimately lies in perception. The questions that demand urgent answers are whether AI’s automation of remix culture threatens to erode the originality and human intent that once defined it and does an AI generated song still count as music?

VI. Conclusion

The fast-paced nature of AI in the creation of music leaves unanswered questions when it comes to copyright law, especially authorship and ownership. The main hypothesis is that there has not been a clear understanding of authorship and copyrightability of a work that is entirely or mostly produced using AI. There is a requirement for amendments to the current legal framework that focuses on clarification under the statute of what constitutes authorship in AI and a sui generis framework with suitability of protections and an increased collective licensing to fairly distribute royalties. It is important that training data be transparent to solve the problem of infringement risks, and international harmonization would ensure international consistency. All these measures are aimed at equating innovation and human creativity and equity in economic participation.

References:

[1] The Copyright Act 1957.

[2] Eastern Book Company v DB Modak (2008) 1 SCC 1.

[3] ANI Media Pvt Ltd v Open AI OPCO LLC, Civil Suit (Commercial) No 1028 of 2024 (Supreme Court, 27 March 2026).

[4] Thaler v Perlmutter 687 F Supp 3d 140 (2023).

[5] The Copyright Act 1957, s 2(d).

[6]  DC Comics v Towle 802 F 3d 1012, 9th Cir 2015.

[7] Copyright, Designs and Patents Act 1988, s 9(3).

[8] Nova Productions Ltd v Mazooma Games Ltd & Ors, [2007] EWCA Civ 219 (CA).

Mimansa Mittal from the Jindal Global Law School, O.P. Jindal Global University, Sonipat, Haryana and Ojas Sharma from the Integrated Law Course (ILC), Faculty of Law, Delhi University, North Delhi