When Algorithms Compose – AI-Generated Music And The Crisis Of Copyright Protection In India
Author: LawSphereX
Category: Intellectual Property
Published:
Updated:
WHEN ALGORITHMS COMPOSE – AI-GENERATED MUSIC
AND THE CRISIS OF COPYRIGHT PROTECTION IN INDIA Ayush Kumar & Manasi Bhardwaj, BA LLB, University School of Law and Legal Studies, GGS Indraprastha University, New Delhi.
ABSTRACT
The rapid proliferation of generative artificial intelligence (GenAI) in the music industry poses an unprecedented threat to the foundational tenets of copyright law and the livelihoods of creators in India. This article examines the multi-faceted crisis emerging from AI-generated music, which mechanically restructures copyrighted datasets without authorization, constituting systemic infringement under Section 51 of the Indian Copyright Act, 1957. It explores how sophisticated voice-cloning technologies erode performers' rights (Sections 38, 38A, and 38B) by misappropriating an artist’s identity without consent, highlighted by recent judicial interventions like Arijit Singh v. Codible Ventures. Furthermore, the paper analyzes the "authorship vacuum," arguing that AI-generated outputs fail the legal "Modicum of Creativity" test established in EBC v. D. B.
Modak and should not be granted copyright ownership. Ultimately, the article advocates for urgent legislative reforms to exclude AI from authorship, mandate licensing for training datasets, and preserve human creativity within an increasingly automated economy. Keywords: Originality, Voice Cloning, Copyright Infringement, Creativity, Authorship, AI-Generated Music, Performers’ Rights, Musicians.
Introduction
It is often said, “Music is food for the soul”, but is music itself losing its soul today? Musicians have always been viewed as the heart and soul of music, as their creativity, skill, and emotional insight give it meaning. However, with the advent of AI, the role of musicians is under fire as AI threatens to replicate the very tasks for which a musician was valued – his voice, composition, and lyrics. As per a recent report, almost 50,000 AI-generated tracks are uploaded to leading music streaming platforms every day. What is worse is that the same study reveals that 97% people cannot distinguish between an original and an AI-generated track. 1 These developments are not merely disruptions, they threaten the very intention and jurisprudence of the copyright law and also raise doubts with respect to the future of musicians.
Copyright is given to protect and promote artistic talent by giving certain exclusive rights and incentives to the author of an original creative work. Copyright rewards the effort put in by the artist to put a melody together, write a song, or sing it. However, in this age, when AI-generated tracks are flooding the internet, the very basis of copyright is struggling to remain relevant. Unlike human beings, AI does not face creative blocks, logistical issues, burnouts, and physical restraints. It can produce an unlimited number of tracks in virtually no time with absolute perfection and precision by learning from both existing copyrighted and non-copyrighted works. Today, AI can mimic, reproduce, and even outperform human artists using the same tools that copyright seeks to protect.
With the current structure, India’s copyright laws are highly unprepared for delivering justice to the musicians left vulnerable due to the AI adoption in this industry. This article examines the ways in which AI-generated music is infringing upon the rights of musicians and creators. It explores how the training of AI systems on copyrighted work may itself constitute a violation of copyright law. Further, the article discusses the growing threat to performers’ rights posed by easily accessible voice-cloning technologies that can replicate an artist’s voice without consent. Finally, it seeks to put forth why AI should neither be granted authorship nor ownership for the music it creates. The article aims to highlight the need for urgent reforms to ensure protection to human artists in an increasingly automated creative economy.
Deezer Newsroom & Ipsos, AI Music Survey, DEEZER NEWSROOM (Nov. 2025), https://newsroom. deezer. com/2025/11/deezer-ipsos-survey-ai-music/.
Infringement of Music Copyright Generative AI models are trained using vast datasets which consist of copyrighted material including lyrics, musical compositions, sound recordings etc. This data is scraped from streaming platforms, music libraries and social media and fed into these AI models without any consent from the rights holders. Section 51 of the Copyright Act states that a copyrighted work shall be deemed to be infringed when any unauthorised person does an act exclusively granted as a right to the copyright owner under section 14 of the Act. 2 Section 14(a)(i) grants a copyright owner an exclusive right to reproduce the work in any material form, including the storing of it in any medium by electronic means. 3 Thus, the act of scraping copyrighted data and feeding it to AI for training purposes without authorization, itself amounts to unauthorized reproduction.
This act does not even fall within the exceptions of this rule given under Section
52 on Fair Dealing because the AI is trained to ultimately produce commercial output.
Unlike human musicians who get inspired by listening to other tracks, AI learns music by copying fully or substantially. AI cannot use creativity to generate something new; as a result, all the content generated by AI is either copied or is a statistical recombination of different copyrighted inputs. Thus essentially, music tracks generated by AI are not unique pieces but an adaptation of a single or multiple copyrighted materials. Making any adaptations to original work is an exclusive right belonging to the owner of the copyright under Section 14(a)(vi). However, it is not possible to identify which works were utilized to generate a particular output because the algorithm's functioning is practically opaque. This opacity makes it almost impossible to protect original works from infringement under the current framework.
The problem with such infringement is that AI is now using protected work to create adaptations or statistical recombinations in absolutely no time and for very little cost. This directly threatens the future of musicians in India because it is now able to create content that cannot be distinguished from original work. A similar case ANI v. OpenAI is currently being heard in the Delhi High Court where ANI has alleged that Chatgpt uses ANI’s content to train its AI model without authorization from ANI. It has also expressed that if such practices continue, it will greatly affect the livelihoods of many young journalists. 4 Similarly, if AI is able to produce unlimited songs using copyrighted material, very soon the population of
2 The Copyright Act 1957 (Act 14 of 1957)
3 Ibid s 14
4 ANI Media Pvt Ltd v Open AI Inc & Anr, CS(COMM) 1028/2024 (Delhi High Court, pending)
musicians, especially the young and emerging artists will be wiped out and be replaced by AI. Voice Cloning And The Erosion Of Performers' Rights One of the most concerning impacts of AI in the music industry is the unauthorised cloning of the voices of popular singers, artists and musicians, making them more vulnerable to misuse and exploitation. The creation of such voice clones requires only a few seconds of the singer’s audio recording to produce near identical results with approximately 95% accuracy. 5 Many well known websites, namely, KITS AI, OpenMusic. AI and ElevenLabs, require only a simple prompt and a few audio samples to generate songs replicating the voices of well-known singers. These fears are not without reason; recently, the Bombay High Court delved into the case of Arijit Singh v.
Codible Ventures6 where the famous playback singer had approached the court to protect his voice and his personality rights against unauthorized use. He sought protection against unauthorised creation of audio and visual content, including mimicry of his voice, mannerisms, vocal style, vocal technique, and vocal arrangements. The use of sophisticated algorithms by generative AI companies for voice cloning makes it difficult to distinguish between the original and AI-created voice. This is alarming because the AI-generated voice sounds almost the same as the original, meaning that it can be used and misused as per the whims of the person generating it, that too without the performers’ consent and with no regard for their intellectual property rights, specifically performers' rights granted under Sections 38, 38A and 38B of the Indian Copyright Act, 1957.
7 These sections bestow upon performers certain special rights known as ‘neighbouring rights’, including the right to record, reproduce and broadcast the performance, the right to collect royalties if their performance is used for commercial purposes, and the right to be identified as a performer and claim damages for any modification of the performance that may be detrimental to the reputation of the performer. A performer's voice is much more than a voice; it is their artistic identity, shaped through years of experience and is an extension of their personality. Unauthorised cloning of this voice violates the rights of a performer and takes away the control over the use of their artistic identity 5 World Intellectual Property Organization (WIPO), AI Voice Cloning: How a Bollywood Veteran Set a Legal Precedent, https://www. wipo.
int/en/web/wipo-magazine/articles/ai-voice-cloning-how-a-bollywood-veteran-seta-legal-precedent-73631
6 2024 SCC OnLine Bom 2445
7 Supra note 4, 1957, §§ 38, 38A, 38B.
and personality. In essence, a performer in this case has no control over what his artistic identity is used to create; as a result, any defamatory, immoral or substandard content generated in the voice of a popular artist will bring bad name and disrepute to the artist with no fault of his. Besides, the biggest attraction for employing such AI tools is their ability to produce premiumsounding results at low cost and low effort. If such a trend continues to gain momentum, it can cause a huge setback to the livelihoods of artists who have invested years of hard work to generate works that are now being used to train AI. Thus granting strong legal protection to an artist’s identity is not only desired, but it is the need of the hour.
Understanding the seriousness of the matter at hand, the Bombay High Court rightly held in favour of Arijit Singh and issued an interim order for protection against any unauthorised use of his voice, vocal style, techniques, and other attributions. However, the gravity of the issue calls for stronger legislation, which warrants prompt legal protection to the performers instead of having to approach the court after their rights have already been infringed. The Authorship Vacuum — Why Ai Cannot Own What It Creates The original intention behind copyright law was not just to protect the work but also to give credit and recognition to the authors or performers for their effort, creativity and originality, none of which is possessed by AI. In the landmark judgment of Eastern Book Company & Ors vs D. B.
Modak & Anr8 the court laid down the Modicum of Creativity test to determine which works can be protected. Thus, only those creations which depict a certain degree of creativity and originality are eligible for protection, whereas mechanical or purely derivative works do not essentially form a part of the subject matter of copyright protection. Music generated by GenAI is not a result of an original creative thought but merely a derivative work based on the data acquired by AI during training, a significant portion of which consists of copyrighted datasets used without prior authorisation of the copyright holder. 9 GenAI mathematically and mechanically restructures the training data to produce an output that appears to be novel on the surface. Human creativity, on the contrary, is shaped by lived experience, emotional depth and analytical judgment, none of which can be reduced to
8 Eastern Book Company & Ors v D B Modak & Anr (2008) 1 SCC 1
9 Department for Promotion of Industry and Internal Trade, Report on Artificial Intelligence and Intellectual Property (DPIIT 2025), https://www. dpiit. gov. in/static/uploads/2025/12/ff266bbeed10c48e3479c941484f3525. pdf. mathematical computation. Furthermore, the present legislation governing the issue, the Copyright Act of 1957, was not even intended to apply to the contents created by the GenAI. Although the 1994 amendment in the Copyright Act mentions that the authorship in the case of computer-generated works goes to the person who causes the work to be done,10 but the provision is ambiguous and fails to address the issue in modern times. It was introduced to address the work generated by early computer programs, which used to require substantial human skill, creativity and labour, unlike the GenAI, which does everything on its own just by simple prompts.
Therefore, the content created by using GenAI does not require any substantial skill, labour, or originality, making it ineligible for the award of copyright. The question of whether AI should be granted copyright cannot be answered without addressing a question: who benefits if AI gets copyright? The copyright is a right in rem, granted to the exclusion of others, conferring several exclusive economic, moral, and neighbouring rights to the copyright holders. Certainly, AI cannot benefit from the copyright granted to it, either economically or morally. Also, the one who causes the GenAI to create certain works cannot be granted copyright since it fails to pass the Modicum of Creativity Test. Therefore, if AI is to be granted copyright, it defaults to the Tech Companies, which may lead to the accumulation of a vast number of copyrights with the tech companies.
A few companies would gain a monopoly over the unlimited data created by AI with minimal effort and negligible cost. In return, the human artists whose data, efforts, skills and experiences have been used for training the AI models will remain at a significant disadvantage or even risk losing their jobs to AI as it becomes more competitive. Ironically, artists may now have to compete against AI models that were once trained on their creative works. The lawmakers never intended to create a monopoly; they intended to protect the skill, labour and originality of the work done by the authors. Consequently, the grant of exclusive rights to an author's work served as an incentive to others and was intended to encourage them to create more for the betterment of society.
Therefore, it is against the interests of society to grant copyright to AI, rather it should be granted only to the original creations that reflect human 10 Zakir Thomas, ‘Overview of Changes to the Indian Copyright Law’ (2012) 17(4) Journal of Intellectual Property Rights 324 https://nopr. niscpr. res. in/bitstream/123456789/14460/1/JIPR%2017%284%29%20324-334. pdf intellect, intellectual effort and human creativity. Conclusion As per a report by IPRS, more than 17,000 musicians are registered under the IPRS as of March
2025.11 Another report by the IPRS and Ernst and Young reveals that there are over 40,000
music creators in India who produce over 20,000 – 25,000 original songs in a year, but not all of them have registered themselves with the IPRS. 12 With AI-generated tracks flooding music streaming platforms, the future of all these musicians, especially the emerging artists, is uncertain. Recently a famous filmmaker in India – Ram Gopal Verma has declared that he will abandon human musicians and use AI generated music in all his upcoming projects. 13 He argues that humans make music production both time consuming and costly while AI delivers instantly and at zero cost. This is exactly what experts have feared. AI is already beginning to replace human artists. AI behaves like a parasite.
It feeds on original copyrighted content from human artists and uses it to mechanically produce thousands of songs at no cost and within no time, stripping musicians of their rights as well as opportunities. The law as it is today is incapable of protecting these musicians and their livelihoods. It was created to protect human work from infringement by humans. The current framework did not anticipate a challenge like the one we face today. As a result, AI is able to defy the purpose of the law and benefit from the existing loopholes. The work or voice of an artist is not just a means of earning livelihood, as rightly held by the Bombay High Court in Arijit Singh v. Codible Ventures, it is a key component of their personal identity and public persona. 14 Thus the law must be reformed to protect human artists before they lose their identity and livelihood to AI.
Section 2(d)(vi) of the Copyright Act must be amended to exclude AI from authorship. Every musical work must satisfy the modicum of creativity test15 and works generated autonomously 11 Indian Performing Rights Society, Annual Report FY 2024–2025 (IPRS 2025), https://iprs. org/wpcontent/uploads/Annual%20Report-FY2024-2025. pdf.
12 Ernst & Young and Indian Performing Rights Society, The Music Creator Economy (IPRS 2024)
https://iprs.org/wp-content/uploads/ey-the-music-creator-economy.pdf
13 Manish Singh, 'Indian Filmmaker Ram Gopal Varma Abandons Human Musicians for AI-Generated Music'
TechCrunch (September 19, 2024)https://techcrunch.com/2024/09/19/indian-filmmaker-ram-gopal-varmaabandons-human-musicians-for-ai-generated-music/ 14 Bombay HC Grants Ad Interim Injunction to Arijit Singh to Protect His Personality Rights, SCC ONLINE BLOG (Aug. 2, 2024), https://www.scconline.com/blog/post/2024/08/02/bomhc-grants-ad-interim-injunction-toarijit-singh-to-protect-his-personality-rights/.
15 Eastern Book Company v D B Modak (2008) 1 SCC 1
by artificial intelligence systems without sufficient human creative input should not be eligible for copyright protection. To stop unauthorised copying of the original work of an artist, AI developers must be required to obtain licences before using copyrighted material for training purposes. Lastly, protections must be to afforded to artists by amending the laws on performers’ rights to prohibit voice cloning. With rapid advancements in AI technology, there is no doubt that AI is more efficient than a human artist. However, musicians have always been the soul of music. It is the emotions and creativity of a musician that create music which touches millions of hearts. The Copyright Act intended to protect this human creativity and effort and today when AI threatens to rob human artists of their identity and recognition, the law must resolve to preserve the same.
The question today is not whether AI will be part of music's future. The question is whether human creators will be.