- within Intellectual Property topic(s)
- within Criminal Law, Environment and Consumer Protection topic(s)
- with Finance and Tax Executives and Inhouse Counsel
- with readers working within the Technology and Law Firm industries
Introduction
The Advent of Artificial intelligence (“AI”) has taken its shape and form in nearly all the sectors, contributing either directly or indirectly to the Indian economy. It has now moved from a topic of revolution to an issue for discussion and more so a reason for restructuring the existing legal regimes. One such sector where AI has settled itself is that of music industry or broadly putting it, the creative industry, amid the growing characterization of the Indian economy as an ‘orange economy,’ driven by creativity, culture, and intellectual property.1
A defining instance of the intersection of AI and the music industry is presented by the establishment and subsequent popularity of AI Music Generator companies, Udio LLC and Suno which are designed to transform text-based prompts or lyrics into fully composed songs, complete with vocals and instrumental arrangements. These AIs utilizes the training corpus of pre-existing songs from publicly available domains, allegedly the ones which are owned by major record labels, namely Sony Music, Universal Music Group (“UMG”) and Warner Records.
Owing to the aforesaid, in the year 2024, Recording Industry Association of America (RIAA) a collective representation from the aforesaid music labels initiated a “mass copyright infringement” proceedings against the AI Music Generator companies, Udio LLC and Suno AI, alleging that these companies have copied hundreds of songs from some of the world's most popular musicians to teach their systems to create music that will "directly compete with, cheapen, and ultimately drown out" human artists”2.
The central issue that has come as a result of these suits, before the Hon’ble Courts of US is that “whether the use of copyrighted songs to train AI constitutes copyright infringement or a “fair use”, a question that US courts have yet to definitively address.”3 While other record labels continue to contest the aforesaid suits, within a year of initiating such proceedings, UMG has opted to settle the matter with Udio LLC, with a view to collaborating on the launch of a platform, expected to be introduced this year, that leverages generative AI trained on authorized and licensed music.”4
Where the aforesaid issue still remains unresolved, this development, encompassing the deal, the infringement proceedings, and, most significantly, the collaboration between record labels and the AI-generating company, has opened a veritable pandora’s box of legal and logistical questions. It has, in turn, placed artists, as the primary owners of copyright, in a position of considerable uncertainty, raising concerns about the potential erosion of their financial interests as well as their relevance and recognition within the creative landscape.
The Copyrightability of AI-Generated Music: A Public Domain Loophole?
A question of fundamental importance, and one that the rise of AI music generators has rendered unavoidable, is whether AI-generated songs are at all capable of attracting copyright protection, and if not, what consequences that implies for the creative economy. The position with respect to the human authorship mandate in most of the countries follows a similar tune.
For instance, the United States maintains a rigid stance on the requirement of human authorship. The U.S. Court of Appeals, in the landmark decision of Thaler v. Perlmutter (decided 18 March 2025)5, the first central case involving copyright in a work produced autonomously by an AI platform categorically held: "Numerous Copyright Act provisions both identify authors as human beings and define 'machines' as tools used by humans in the creative process rather than as creators themselves. Because many of the Copyright Act's provisions make sense only if an author is a human being, the best reading of the Copyright Act is that human authorship is required for registration."6
India also takes a position similar to that of the U.S. but is distinguished by the absence of any binding Supreme Court or High Court precedent specifically addressing autonomous AI creation. However, as recently as in June 2026 the Copyright Office of India, fortifying its position, updated its Form XIV application framework, requiring applicants of AI-assisted works to proactively upload clear documentation demonstrating "substantial human creative direction, selection, and arrangement" to establish legal authorship, aligning Indian administrative practice with these global standards.7
The Loophole: If AI-generated music does not qualify for copyright under either the U.S. or the Indian framework, the logical consequence is that such music falls immediately into the public domain. This creates a significant and under-examined loophole: AI companies may generate vast quantities of music derivative of human artistic works, yet because those AI outputs attract no copyright, they may be freely reproduced, distributed, and commercially exploited by any party. This paradox — where AI is trained on protected human works but itself generates unprotected works, simultaneously undermines the incentive structure of copyright law and places human creators at a structural competitive disadvantage. In the music industry, this is particularly concerning: AI-generated songs could flood streaming platforms as public domain content, diminishing the market for human-authored compositions without attracting any obligations of attribution or remuneration.A Worldwide view on the debate of “Fair Use” v. AI utilizing Copyrighted Works
The use of AI to process, reproduce, and disseminate music and other original works has become an increasingly pervasive phenomenon within the expanding contours of copyright law worldwide. Notably, unlike India, where the debate is still at a relatively nascent stage, this issue has already assumed center stage in several foreign jurisdictions. While no comprehensive or express legislative framework has yet been enacted to regulate AI training on copyrighted works, courts abroad have proactively engaged with the issue by interpreting the contours of the “fair use” doctrine in the context of AI-generated content.
A significant illustration of this development can be found in Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc.8, adjudicated by the U.S. District Court for the District of Delaware. The dispute arose from Ross Intelligence’s use of copyrighted legal headnotes and the Key Number System owned by Thomson Reuters (Westlaw) to train its AI-driven legal research platform. In rejecting the defense of fair use, the Court held that Ross’s use was commercial in nature and lacked trans- formativeness, as the AI tool served a substantially similar purpose to Westlaw’s proprietary offerings, namely, facilitating legal research. The Court further observed that such use adversely impacted the market of the copyright holder. This decision is widely regarded as one of the earliest judicial pronouncements directly limiting the applicability of fair use in the context of AI training, and its implications extend to other creative domains, including music and literary works.
At the same time, judicial approaches have not been entirely uniform. In Kadrey v. Meta Platforms, Inc.9, before the U.S. District Court for the Northern District of California, the Court granted summary judgment in favour of Meta on key copyright claims concerning AI training. Central to the ruling was the plaintiffs’ “market dilution” theory — that AI-generated output would flood the market with similar works and depress demand for the originals — which the Court found unsupported by any evidence in the record. The Court nonetheless cautioned that a properly evidenced market-dilution claim could decisively tip the fourth fair-use factor against an AI developer in a future case; the ruling itself binds only the thirteen named plaintiffs, not a certified class of authors.
Thus, while foreign courts are actively shaping the jurisprudence by attempting to strike a balance between technological innovation and the protection of authors’ rights, a consistent and settled legal position is yet to emerge. Nevertheless, these developments underscore a clear trend: courts are increasingly inclined to scrutinise the application of fair use in AI contexts through the lens of market impact, trans-formativeness, and commercial intent, an approach that may offer valuable guidance as Indian jurisprudence begins to engage more substantively with similar questions.
Right to Integrity: A challenge beyond the dilemma of fair use
The challenge posed by AI to the creative industry extends beyond copyright in training data. The proliferation of AI-generated deepfake vocals and voice-cloning technology strikes directly at an author's moral rights, rights which, unlike economic rights, cannot be transferred and which survive the assignment of copyright itself.
Section 57 of the Indian Copyright Act, 1957 enshrines an author's Right to Integrity, permitting them to claim damages for any distortion, mutilation, or modification prejudicial to their honour or reputation. The Delhi High Court in Amarnath Sehgal v. Union of India10 affirmed that moral rights, encompassing the Paternity, Integrity, and Retraction Rights, form "the soul" of an author's work, while Raj Rewal v. Union of India11 in fact reached the opposite conclusion on these facts: the Court held that “distortion, mutilation, or modification” under Section 57(1)(b) reaches only partial alteration of a work, so the complete destruction of the Hall of Nations fell outside its scope — a limitation reinforced by the architect’s statutory right yielding to the Union’s constitutional property right under Article 300A. The provision’s protection against partial distortion, however, remains undisturbed.
Although India lacks dedicated AI/deepfake legislation, courts have increasingly applied these principles to AI-generated content. In Arijit Singh v. Codible Ventures LLP12 (Bombay HC, 2024), the Court extended personality rights to vocal styles and AI voice-cloning, while in Kumar Sanu Bhattacharjee v. Jammable Limited13 (Delhi HC, 2025), it directed takedowns across platforms and ordered MeitY and DoT to suspend infringing websites — reaffirming that AI imitations of celebrity personas breach moral and creative ownership. Accordingly, where an AI-generated work employs a cloned voice in a poor-quality, objectionable, or reputation-damaging manner, it is capable of constituting a violation of Section 57.
Therefore, while deciding the aspect of “fair use”, protecting the right to integrity of the author, shall also form one of the fundamental consideration.
In summation: A dilemma that awaits for the copyright holders
The rise of AI in the creative industry is not a distant disruption, it is a present commercial reality reshaping the foundations of copyright, royalty entitlement, and moral rights. The collaboration between UMG and Udio LLC, the infringement proceedings by RIAA, and the evolving settlement landscape collectively demonstrate that industry practice is racing ahead of legal clarity. India's copyright regime, as amended in 2012 and 2023, provides a foundational structure protecting both underlying works and sound recordings, but does not adequately address the specific challenges posed by AI training on copyrighted data, the copyrightability of AI-generated music, the violation of moral rights through deepfake synthesis, or the mechanisms needed to ensure fair remuneration to human creators in the AI era.
The resolution of ANI v. OpenAI by the Delhi High Court will be one of the most consequential determinations in the history of Indian copyright law, framing the permissible scope of AI training on copyrighted data for decades to come. Yet judicial pronouncements alone, necessarily retrospective and case-specific, cannot provide the systematic framework that India's creative economy urgently requires. The reactive but jurisdiction-limited protections extended in Arijit Singh, Aishwarya Rai Bachchan, and Kumar Sanu underscore this structural gap: courts have been willing, but legislation has been absent.
India must therefore act with purposeful deliberation. A qualified Text and Data Mining exception, modelled on the EU Directive 2019/790 but calibrated to Indian realities, would resolve the present uncertainty under Section 52 while preserving rights-holders' opt-out prerogatives. This should be complemented by a compulsory licensing regime for commercial AI training, enabling collecting societies such as IPRS to administer royalties systematically, and by a Centralised AI Training Dataset Registry requiring disclosure of training corpora to support both enforcement and transparency. Finally, dedicated legislation prohibiting the unauthorised commercial use of an artist's voice, image, or likeness through AI synthesis, with meaningful civil and criminal remedies, is essential to bridge the gap that courts have been filling piecemeal.
The challenge, as this article recognises, is not without precedent; the law has navigated transformative technological disruptions before, from sound recording to digital broadcasting. But it remains critically under-examined in the Indian context. As platform collaborations between music labels and AI companies multiply, potentially reshaping the practical meaning of fair use before courts or legislation have settled the question, the time for that systematic examination is unequivocally now. A structured legal framework is not merely desirable; it is owed to the human creators whose labour, artistry, and reputation underpin the very works that AI systems are trained upon.
Footnotes
4 Supra.
5 https://media.cadc.uscourts.gov/opinions/docs/2025/03/23-5233.pdf
6 Supra, 9
8 Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc., No. 1:20-cv-00613-SB, 2025 WL 458520 (D. Del. Feb. 11, 2025) (revising in relevant part 694 F. Supp. 3d 467 (D. Del. 2023)).
9 Kadrey v. Meta Platforms, Inc., No. 3:23-cv-03417-VC, (N.D. Cal. June 25, 2025).
10 2005 (30) PTC 253 (DEL)
11 CS (COMM) 3/2018, IA No. 90/2018, IA No. 92/2018
12 2024 SCC OnLine Bom 2445
13 CS(COMM) 1097/2025, Delhi High Court, decided on 15 October 2025.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
