Arijit Singh v. Codible Ventures: AI Voice Cloning and Personality Rights in India

Contributed by: Priya Dutt; Reviewed by Adv. Manvee

Introduction

The Bombay High Court’s ex parte ad interim order in Arijit Singh v. Codible Ventures LLP and Ors. is a significant decision on personality rights, publicity rights, AI voice cloning, deepfakes, and unauthorized commercial use of a celebrity’s identity in India. This case analysis explains the facts, issues, reasoning, interim relief, and practical implications for artists, AI platforms, creators, and brands in India.

Arijit’s case represents a significant development in India’s evolving personality rights jurisprudence, particularly adapted to confront the advanced challenges of generative artificial intelligence, voice cloning, and deepfake synthesis. Arijit Singh, a renowned playback singer and personality, is a global icon whose legal struggle represents a seminal case for the legal protection of an artist’s digital personality in a world where technology is often in a race with current legislation.

The nature of the dispute in Arijit Singh v. Codible Ventures is the unlicensed commercial exploitation of the plaintiff’s most intimate and distinctive personality traits in the form of his voice, vocal style, mannerisms, and performance persona. The defendants in this case are AI platform owners and creators of artificial media who have used sophisticated algorithms to reproduce the plaintiff’s distinctive personality traits without his permission or license and have essentially commercialised the plaintiff’s identity for their own financial gain.  

It is particularly noteworthy that Arijit’s case marks the first instance in Indian law dealing specifically with the misappropriation of generative AI technology in relation to the violation of personality rights. It challenges the status quo in intellectual property and rights law by arguing that a person’s persona, especially a unique voice, is a form of intellectual property that cannot be exploited by technological ingenuity. The case represents a clash between technological ingenuity and a person’s inherent right to control the commercial use of one’s identity, which is increasingly being accepted as a significant aspect of a person’s dignity and economic value.  

Quick Summary of the Arijit Singh AI Voice Cloning Case

• Court: Bombay High Court
• Judge: Justice R.I. Chagla
• Date of Order: 26 July 2024
• Case Type: Commercial IPR suit / interim application
• Main Issue: Unauthorized use of Arijit Singh’s name, voice, image, likeness, persona, and other personality traits
• Key Technology Involved: AI voice cloning, generative AI, deepfakes, GIFs, and digital platforms
• Relief Granted: Ex parte ad-interim injunction, including dynamic injunction-style relief

Facts of the Case

The case involved 38 defendants, including AI platform operators, e-commerce sellers, domain name registrars, cloud/data storage platforms, and unknown “John Doe/Ashok Kumar” defendants. The Plaintiff (inter alia “Arijit Singh”) alleged that these defendants misused his personality traits, including his name, voice, vocal style, image, likeness, signature, and persona, without authorization. The order records allegations that certain AI platforms enabled users to convert speech, text, or audio into an AI-generated version of Arijit Singh’s voice, including through Real Voice Cloning technology, and that a dataset of 456 songs from the Plaintiff’s repertoire had been uploaded without authorisation for voice-conversion purposes. The Plaintiff also alleged that the founder of Defendant No. 2 promoted tutorials showing users how to clone Singh’s voice using the relevant AI platform.

Key Allegations Before the Bombay High Court

The Plaintiff alleged the following forms of unauthorized use:

1. AI voice cloning: AI platforms allegedly enabled users to generate audio in Arijit Singh’s voice.
2. Use of 456 songs: A dataset of 456 songs from the Plaintiff’s repertoire was allegedly uploaded to enable voice conversion.
3. YouTube tutorials: Certain videos allegedly gave step-by-step guidance on converting voice/audio into Arijit Singh’s AI voice.
4. False association: Events and platforms allegedly used his name and image to suggest endorsement or participation.
5. Unauthorized merchandise: Products such as clothing, mugs, phone cases, masks, posters, and other items allegedly used his name/image/likeness.
6. GIF misuse: Platforms allegedly enabled the creation and sharing of GIFs using his performances and image.
7. Domain name misuse: Domain names such as arijitsingh.com and arijitsingh.in were allegedly registered without authorization.

Beyond AI voice cloning, the Plaintiff alleged broader commercial exploitation of his publicity rights through false association, unauthorised merchandise, GIFs, and domain-name misuse. The Plaintiff’s case was that these activities were not isolated fan uses, but a pattern of unauthorised exploitation of goodwill and celebrity identity that could affect his reputation, endorsement value, and livelihood as a professional singer.

Legal Issues in Arijit Singh v. Codible Ventures

The Bombay High Court considered whether unauthorized use of a celebrity’s name, voice, likeness, persona, image, signature, and singing style can amount to a violation of personality rights and publicity rights.

The key legal issues were:

1. Whether AI voice cloning without consent violates personality rights.
2. Whether unauthorized commercial use of a celebrity’s persona violates publicity rights.
3. Whether the Plaintiff’s moral rights as a performer under Section 38-B of the Copyright Act 1957 were implicated.
4. Whether videos/tutorials enabling AI voice replication can be protected under freedom of speech and expression.
5. Whether ex parte ad-interim relief and dynamic injunction-style relief were justified.

Arguments by the Parties

Plaintiff’s Contentions

The legal arguments for Arijit Singh were based on the school of thought that an individual’s identity is an asset that has been cultivated with the hard work and dedication of several years. The plaintiff argued that an individual has an absolute right to command and control the commercial use of his personality traits since they are the primary identifiers of the individual’s brand and source of livelihood. 

The plaintiff argued that the use of voice cloning is an extremely invasive form of technological exploitation. The defendant’s use of AI technology and training it with a dataset of 456 unauthorised songs was not merely replicating the sounds but was essentially “pirating” the essence of Singh’s professional identity. The plaintiff argued that this violated the exclusive right of an individual to commercially exploit his personality and also amounted to the tort of dilution by tarnishment.

Moreover, it was contended that “unauthorized transformation or synthesis of the plaintiff’s voice creates deep prejudice and injury to the plaintiff’s professional honor and thus violates the statutory moral rights of the plaintiff as a performer under Section 38-B of the Copyright Act.” The plaintiff’s counsel also highlighted his “conscious personal choice” of not engaging in gross commercialization and brand endorsements for several years and thus the forced engagement by the defendant had an even more damaging impact on the plaintiff’s reputation.  

Potential Defences/Issues Not Fully Tested at the Ex Parte Stage

Because the order was passed ex parte, the defendants did not substantively present their defences at this stage. Any discussion of fair use, transformative use, parody, education, safe harbour, or the absence of a dedicated AI statute should therefore be presented as potential issues for later adjudication, not as actual defendant submissions in this order.

Court’s Reasoning

Justice R.I. Chagla’s reasoning focused on three points: first, celebrity personality attributes are protectable against unauthorised commercial exploitation; second, AI tools that enable conversion of a person’s voice into a celebrity’s voice can facilitate appropriation of a key identity marker; and third, urgent ex parte relief was justified because online and AI-enabled misuse can spread quickly and cause irreparable harm.

(a) Recognition of Personality Rights

The court started by holding that celebrities have a right to a high degree of protection for their “personality rights,” which include name, image, likeness, voice, and signatures. This right is based upon a “right to privacy.” The Supreme Court of India has held that this right is a fundamental right under Article 21 in Puttaswamy vs Union of India.  The Court reached a prima facie view at the ad-interim stage and relied on existing celebrity-rights precedents, including Karan Johar, Anil Kapoor, Amitabh Bachchan, D.M. Entertainment, and Rajagopal as discussed in the order.

The rationale for this decision is that a person’s identity is not just a matter of personal information but is a form of economic interest that a person has a right to control and that allowing others to use a celebrity’s identity would be tantamount to allowing others to “ride upon” a celebrity’s hard-earned fame and success. The court noted that fame subjects a person to a variety of vulnerabilities, such as the theft of a celebrity’s endorsement value, which is a significant source of living.  

(b) Voice and Likeness as Protectable Attributes

An important aspect of the reasoning was the court’s careful consideration of the “protectable facets” of Singh’s personality. The judgment also determined that the voice of a professional singer is as “distinctive and personal as a face”. 

The court argued that the imitation of a singer’s voice by AI synthesis amounts to “AI voice cloning may amount to unauthorised appropriation of identity”. This is because the singer manifests herself in her song, and the tool that imitates her voice without her permission takes away her ability to control her own identity.

(c) Digital Misappropriation and Technology

Justice Chagla’s profound “shock to the conscience” was directed towards the targeting of performers by generative AI content. The court noted how AI platforms were “leveraging” Arijit Singh’s popularity to attract traffic and “embolden” users to create “counterfeit” recordings. 

The Court was particularly concerned that AI-generated voice and video content could be misused for commercial, deceptive, or nefarious purposes. It also clarified that freedom of speech may protect critique and commentary, but does not permit commercial exploitation of a celebrity’s persona without consent.

(d) Need for Injunctive Relief

In arriving at its decision to grant ad-interim relief, the court had to apply the threefold test that is usually applicable while granting injunctions: the prima facie case, balance of convenience, and irreparable injury. The court found that the documents filed were sufficient to prove that Singh was a celebrity and that the misuses by the defendants were for pecuniary gain.

The balance of convenience was clearly in favor of the plaintiff, as the stakes were high, not only for his career but also because, in the age of instantaneous dissemination of content, monetary damages would not have been sufficient. This called for a flexible injunction that could reflect the fluidity of the internet space and mirror sites. 

Order and Relief Granted

The Bombay High Court granted an ex parte ad-interim injunction. For accuracy, do not call this a final decision or final judgment. The order restrained unauthorised use of Singh’s personality attributes and directed takedown, domain-locking/suspension, disclosure, and access-disabling measures against the relevant defendants and intermediaries.

The operative part of the order contained the following directions:   

1. Restraint on Persona Use: The defendants, as well as all the associated entities, were restrained from using Arijit’s name, voice, picture, image, caricature, or signature for any commercial or personal benefit without his express consent.

2. AI Prohibitions: The court specifically prohibited the development or use of AI voice models, synthetic voice, avatar, as well as deepfakes that simulate, mimic, or replicate Arijit’s personality traits.

3. Content Takedown and Domain Suspension: The defendants were ordered to take down the infringing content, including tutorial videos as well as song covers. The Domain Name Registrars (DNRs) were ordered to lock and suspend the infringing domain names, such as arijitsingh.com, with a prohibition on the transfer of the domain names to third parties.

4. Regulatory Interventions: The Department of Telecommunications (DoT), as well as the Ministry of Electronics and Information Technology (MeitY) were ordered to direct all Internet Service Providers (ISPs) to disable access to the particular infringing URLs.

5. Disclosure Mandates: E-commerce as well as cloud storage service providers, including Amazon, Flipkart, and Google, were ordered to disclose all the details of the defendants involved in the unauthorised sale of merchandise.

The court emphasised that while freedom of speech permits “critique and commentary,” it does not grant a license to exploit a celebrity’s persona for commercial profit.   

Key Legal Principles from this Case

The Arijit Singh judgment has distilled several core principles that serve as the bedrock for the future of digital identity law in India :   

• Vocal Distinctiveness as Identity: The case conclusively determines that the vocal characteristics, mannerisms, and style of a professional are integral parts of the persona, which are protectable as a component of both personality rights and publicity rights.
• Liability for Technology Providers: In holding the platform operators liable, the court opined that the provision of tools for voice cloning, which enables identity theft, is actionable, irrespective of the party completing the process.
• The Consent Threshold: Consent is recognised as the “gateway requirement” for the use of an individual’s personal characteristics for commercial purposes. The judgment asserts that the unauthorised use of an artist’s body of work for AI model development violates both the economic and moral rights of the artist.
• Functionalist Property Rights: The judgment recognises the “quasi-property right” of personality rights, which reflects the functionalist approach, balancing dignity-based privacy with the proprietary interest in fame.
• Moral Rights in the Synthetic Era: The case applied Section 38-B of the Copyright Act to the synthetic media context, affirming that AI replicas can be prejudicial to a performer’s honour and reputation, thus triggering statutory moral protections.   

Critical Analysis

The order is an important development, but the analysis should remain measured. It strengthens personality-rights protection against AI-enabled misuse, while also exposing unresolved questions on statutory codification, intermediary liability, fan works, parody, training data, and non-celebrity voice protection.

Expansion of Personality Rights Jurisprudence

The order broadens the practical understanding of likeness and persona by including voice, vocal style, and mannerisms. However, avoid inventing labels such as ‘commercial gestalt’ unless you define them as commentary. The more defensible analysis is that Indian courts are moving toward consent-based control over commercially valuable identity attributes.

Lack of a Statutory Framework

India still lacks a single dedicated statute on personality rights or AI voice cloning. Claimants currently rely on a combination of personality/publicity rights, privacy principles, passing off, copyright/performance rights, tort principles, and injunctive relief. This patchwork creates uncertainty, especially for non-celebrities and for lawful commentary, parody, and fan expression. This has resulted in a lack of consistency and has led to the application of different tests by different High Courts. For example, the test applied by the Delhi High Court is the “Likelihood of Confusion” test, while the test applied by the Bombay High Court is the “Commercial Identification” test.

Judicial Approach to Emerging Technologies

The intervention of the court in editing specific content in YouTube videos instead of a blanket ban on channels shows a high degree of understanding and awareness about balancing rights. However, the continued availability of cloned content on some platforms in spite of this order shows that courts move at a “glacial” pace in contrast to the “instantaneous” pace of cybercrimes and cyberlaws.

How the Arijit Singh Case Shapes AI Voice Cloning and Deepfake Regulation

(a) Impact on AI Voice Cloning Tools

AI voice-cloning and text-to-speech platforms should treat this order as a compliance warning. Practical controls include consent checks for celebrity voice models, notice-and-takedown workflows, repeat-infringer controls, dataset provenance review and filters for protected names and voices.

(b) Relevance for Generative AI Platforms

Generative AI platforms should adopt transparency-by-design, including clear AI-generated labels, provenance metadata where feasible, and user-facing restrictions on cloning identifiable voices without consent. The Arijit Singh case underscores that fair use/free speech arguments are weaker where the use is commercial, misleading or identity-exploitative.

(c) Implications for Influencer & Creator Economy

For influencers and creators, the case signals that distinctive identity features – including voice, image, catchphrases, delivery style, and performance persona – may have legal and commercial value. Not every meme, parody, critique, or fan edit will automatically be unlawful; context, purpose, identifiability, and commercial use matter.

(d) Lessons for Tech Startups and Platforms

For startups and platforms, the practical legal takeaway is to build a consent-and-risk framework before launching voice-cloning, avatar, deepfake, or celebrity-style AI features. Key controls include:

• Explicit consent and licensing: obtain documented permission before offering, training, or monetising identifiable celebrity voice or likeness models.
• Dataset provenance: maintain records showing that training and reference material is lawfully sourced and permitted for the intended use.
• Synthetic disclosure and user controls: label AI-generated outputs, prohibit deceptive celebrity impersonation, and create rapid takedown channels for rights holders.

Comparison with Global Deepfake and Voice Rights Cases

US Precedents: Bette Midler and Tom Waits

The judicial reasoning in India mirrors foundational cases from the United States. In Midler v. Ford Motor Co. (1988), the 9th Circuit established that “We hold only that when a distinctive voice of a professional singer is widely known and is deliberately imitated in order to sell a product, the sellers have appropriated what is not theirs and have committed a tort in California”.[1] Similarly, in Waits v. Frito-Lay, Inc. (1992), the court protected Tom Waits’ raspy voice against a deliberate sound-alike imitation used for a commercial, awarding him $2.6 million in damages.  The Ninth Circuit Appellate Court found that when voice is a sufficient indicator of a celebrity’s identity, the right of publicity protects against its imitation for commercial purposes without the celebrity’s consent. This right is protected in California understate tort law. Under the federal Lanham Act, celebrity singers have standing to sue for voice misappropriation because of the likelihood that wrongful use of their professional trademark-their voices-would injure them commercially.[2]

Unlike the US cases, which focused on human mimics, the Arijit Singh case addresses the massive scalability of AI replication, which can generate thousands of unauthorised performances in seconds.   

Statutory Trends: Tennessee ELVIS Act and EU AI Act

Jurisdictions are increasingly recognising voice and likeness as protectable identity attributes, particularly in response to AI-generated impersonation. Tennessee’s ELVIS Act (2024) specifically adds “voice” as a protectable attribute to its publicity statutes, aimed directly at AI companies. The Act defines “voice” to include “a sound in a medium that is readily identifiable and attributable to a particular individual, regardless of whether the sound contains the actual voice or a simulation of the voice of the individual.” This broad definition raises the spectre that liability under the ELVIS Act may extend not only to the use of an existing sound recording of someone’s voice, and not only to digitally generated recordings or audiovisual content that approximates individual voices, but also to humans who can imitate other artists (i.e., soundalike artists).[3]

The obligations for disclosing synthetic content, including deepfakes, primarily fall under Article 52 (transparency requirements for general-purpose AI models) and related Title III provisions for high-risk systems. While Article 5 prohibits certain manipulative AI practices, it does not specifically target deepfakes; violations can incur fines up to 7% of global turnover under Article 101. 

[4]   

FAQ: Arijit Singh v. Codible Ventures and AI Voice Cloning in India

What did the Bombay High Court decide in Arijit Singh v. Codible Ventures?

The Court granted ex parte ad-interim relief restraining unauthorised use of Arijit Singh’s name, voice, likeness, image, signature, persona and other personality attributes, including through AI voice models and deepfakes.

Is AI voice cloning illegal in India?

India does not yet have a standalone AI voice-cloning statute. However, unauthorised AI cloning of an identifiable celebrity’s voice for commercial or deceptive use may be restrained through personality rights, publicity rights, privacy principles, copyright/performance rights and injunctive relief.

Why is the case important for AI platforms?

The order indicates that platforms offering tools to convert voices into a celebrity’s voice without consent may face injunctions, takedown directions, and disclosure obligations, especially where the use attracts traffic or commercial gain.

Was this a final judgment?

No. The order was an ex parte ad-interim order.

What compliance steps should startups take?

Startups should obtain consent for identifiable voice or likeness models, document dataset provenance, label synthetic content, prohibit deceptive impersonation and maintain rapid takedown channels.

Conclusion

Arijit Singh v. Codible Ventures is an important interim order for personality rights and AI voice cloning in India. It signals that courts may restrain unauthorised AI voice models, deepfakes, and commercial uses of a celebrity’s persona where the plaintiff is identifiable, consent is absent, and reputational or economic harm is likely.

The broader significance is that Indian personality-rights law is being tested against synthetic media. The order supports stronger consent, disclosure, and takedown practices, but it also shows the need for clearer statutory rules on AI voice cloning, training data, platform responsibility, and the limits of lawful commentary, parody, and fan content.

AI innovation is not barred, but AI products that monetise or simulate identifiable voices and personas should be built around consent, labelling, provenance and rapid redress.

[1] https://cyber.harvard.edu/people/tfisher/1988%20Midler.pdf

[2] https://via.library.depaul.edu/cgi/viewcontent.cgi?article=1489&context=jatip

[3] https://www.lw.com/admin/upload/SiteAttachments/The-ELVIS-Act-Tennessee-Shakes-Up-Its-Right-of-Publicity-Law-and-Takes-On-Generative-AI.pdf

[4] https://artificialintelligenceact.eu/article/5/