Rethinking Performers’ Rights for AI-Generated Digital Personas

Introduction : A performer’s voice has traditionally been inseparable from the performance in which it is expressed. An actor speaks a dialogue, a singer records a song, and the law can identify a concrete performance, a recording and the economic rights attached to them. Artificial intelligence disrupts that relationship. A voice can now be extracted as a recognisable personal attribute, analysed through existing recordings and reproduced in an entirely new performance that the individual never gave. The resulting audio may sound authentic even though there is no corresponding human performance behind it.

This creates a legal distinction that is easy to overlook. The issue is not necessarily whether an existing song or recording has been copied. It is whether the identity of the performer has itself been reproduced in synthetic form. That distinction becomes particularly significant after the performer’s death. An estate may possess copyright interests in recordings or contractual rights connected with earlier works, yet the synthetic voice may be a newly generated output rather than a copy of any one protected recording.

India’s Copyright Act, 1957 already recognises performers as holders of a distinct category of related rights. A performer includes an actor, singer, musician and other persons who make a performance. Yet the statutory structure was developed around performances rather than the broader idea of a digital persona. The emergence of voice cloning therefore exposes a question that copyright law did not originally have to answer: when technology can recreate the performer without reproducing the performance, what exactly is the law protecting?

What India’s Copyright Act protects, and where the protection becomes uncertain

Sections 38, 38A and 38B of the Copyright Act form the statutory core of performers’ rights. Section 38 creates a special right in relation to a performance, but limits its duration to fifty years from the beginning of the calendar year following the year in which the performance is made. Section 38A gives performers exclusive rights in relation to the recording and communication of their performances, including reproduction of a sound or visual recording and communication to the public.

The architecture matters because these rights are connected to a particular performance. A straightforward case involving the unauthorised copying of an existing recording therefore fits comfortably within copyright doctrine. A synthetic song, however, may be different. Suppose an AI system is trained on numerous recordings of a singer and subsequently generates a new track in a voice that is recognisably that singer’s, without reproducing a substantial portion of any original recording. The economic value being appropriated is the singer’s vocal identity, not necessarily the protected fixation itself.

Section 38B introduces a further layer by granting performers moral rights. A performer may claim identification and may restrain or seek damages for distortion, mutilation or other modification of the performance that is prejudicial to reputation. This provision is potentially important for AI because synthetic manipulation can alter what an audience believes the performer actually said or sang. Yet the wording still centres on modification of a performance. A wholly new AI-generated performance that merely imitates the performer may fall outside the most natural reading of that provision.

The limitations are also visible in section 39, which preserves specified uses from infringement, including fair dealing with excerpts of performances for reporting current events, review, teaching and research. The problem is therefore not that the Copyright Act has no safeguards. It is that those safeguards were designed to regulate the exploitation of performances as protected subject matter, whereas generative AI can exploit the performer’s identity as a source from which a new synthetic performance is produced.

The distinction becomes sharper when compared with section 57, which deals with authors’ moral rights. The Act expressly permits the legal representatives of an author to exercise the rights under section 57. Section 38B contains no comparable express provision dealing with the posthumous exercise of performers’ moral rights. That difference is legally significant because it is precisely after death that a dispute over an AI-resurrected voice may arise, while the person whose identity was appropriated can no longer personally consent or object.

Indian courts are protecting voices, but mainly through personality rights

Indian courts have increasingly addressed the precise harm that the Copyright Act does not squarely describe. In Anil Kapoor v. Simply Life India, the Delhi High Court protected a wide range of attributes, including the actor’s name, image, likeness, voice and manner of speaking, and restrained unauthorised commercial exploitation through technologies including artificial intelligence and deepfakes. The significance of the decision lies in its recognition that identity can possess independent legal and commercial value even when the dispute does not depend entirely on copying a copyrighted work.

The Bombay High Court moved further in Arijit Singh v. Codible Ventures LLP. The singer sought protection not merely for existing recordings but for his voice, vocal style, vocal technique, manner of singing and other features of his persona. The Court granted interim protection against the unauthorised use of AI tools and other technologies to replicate those attributes. The reasoning demonstrates why personality rights have become particularly important in the AI context. A voice-cloning platform may not need to reproduce a particular song in order to commercially exploit what makes that singer identifiable.

The same concern appeared in Asha Bhosle v. Mayk Inc., where the Bombay High Court granted interim protection against the use of AI tools capable of converting another voice into Bhosle’s voice without consent. The order treated the voice as an important component of personal identity and recognised the risks associated with commercial and deceptive exploitation of that identity. These decisions collectively show that Indian courts are capable of responding to synthetic identity appropriation even without a dedicated statutory digital-replica right.

But judicial development has an important boundary. Personality rights in India do not presently rest on one comprehensive legislation. More importantly, their posthumous status remains unsettled. In Krishna Kishore Singh v. Sarla A Saraogi, concerning the use of Sushant Singh Rajput’s persona, the Delhi High Court held that personality rights, if any, did not devolve on his father after his death. The Court treated the relevant right as personal rather than heritable. Although the factual setting involved a film rather than an AI voice clone, the principle presents a direct difficulty for digital resurrection. An estate cannot necessarily assume that the legal control available to a living performer automatically survives death.

That creates an unusual mismatch. Technology allows the commercial recreation of a deceased performer with increasing ease, while the legal system may protect the recordings left behind without clearly protecting the performer’s identity as such. The gap is not merely about compensation. It concerns who has authority to decide whether a deceased person may appear to sing a new song, endorse a product, deliver new dialogue or be placed in an entirely new creative setting.

The comparative shift from existing copyright to digital replica rights

The United States has moved the debate towards a more explicit statutory model. The U.S. Copyright Office, after examining digital replicas, concluded that existing law leaves important gaps and recommended a new federal right against the knowing distribution of unauthorised digital replicas, with licensing, monetary and injunctive remedies and safeguards for constitutionally protected expression.

The proposed NO FAKES Act of 2025 illustrates what such a right could look like. Introduced in both chambers, the proposal defines a digital replica as a highly realistic computer-generated representation of an individual’s voice or visual likeness and creates a property right controlling its authorised use. Its treatment of deceased individuals is particularly relevant to India. The proposed right would survive death, initially for ten years, with further five-year renewals based on continued authorised public use, subject ultimately to a seventy-year postmortem ceiling. This is materially different from relying on copyright in an old recording or attempting to stretch a personality right through judicial interpretation.

The European Union illustrates a different regulatory approach. Its AI Act does not establish a general property right in a person’s voice or likeness. Instead, Article 50 imposes transparency obligations for AI-generated and manipulated content, including disclosure of deepfakes, with the relevant transparency obligations applying from 2 August 2026. The European Commission also stated in 2025 that it did not plan, at that stage, to introduce specific regulation of automated AI dubbing, while noting that existing EU copyright already gives performers rights over the fixation of their performances.

The comparison is revealing because there is no single global model emerging. The US debate centres on control over the digital replica itself. The EU places greater emphasis on transparency and the ability of audiences to identify synthetic content. India, meanwhile, is developing protection largely through a combination of statutory performers’ rights and judicially recognised personality and publicity rights.

What India should protect

The answer is not necessarily to convert every aspect of human identity into copyright. Copyright exists to protect defined creative subject matter, and extending it indiscriminately to a person’s voice, face or mannerisms could distort the conceptual boundaries of intellectual property. A new statutory protection would therefore need to distinguish between a protected performance and the performer’s digital persona.

The more coherent approach would be to strengthen the performers’ rights framework with an express digital-replica provision, while preserving personality rights as an independent basis of protection. Such a provision could define an unauthorised digital replica broadly enough to cover realistic synthetic reproductions of a performer’s voice or likeness, without requiring proof that an existing recording was substantially copied.

Consent should be central. A performer should be able to authorise specified uses of a digital replica through an informed written agreement, rather than through vague contractual language drafted for conventional exploitation. The law should also address what happens after death. A limited postmortem right, administered by clearly identified legal representatives or another designated rights holder, would provide certainty without creating perpetual control over a cultural figure.

At the same time, exceptions are essential. Legitimate biography, documentary, criticism, parody, scholarship, historical reconstruction and other protected forms of expression cannot be treated as commercial appropriation merely because they involve a recognisable person. The US proposal and the EU framework both demonstrate, in different ways, that identity protection has to coexist with freedom of expression rather than displace it.

India’s recent amendment of the Information Technology Rules also shows that the regulatory system is already moving towards labelling and platform accountability for synthetically generated information. The 2026 framework introduces requirements relating to labelling, traceable metadata and faster removal of unlawful synthetic content. These measures can assist enforcement, but they do not answer the prior question of who owns or controls a deceased performer’s synthetic identity.

Conclusion

AI has changed the legal object at stake. The difficult case is no longer simply the unauthorised copying of a song, film or recording. It is the creation of something new that nevertheless derives its commercial value from making the audience believe that a particular human performer is speaking or singing.

India’s courts have shown that existing personality-rights doctrine can respond to this problem in individual cases, as Anil Kapoor, Arijit Singh and Asha Bhosle demonstrate. Yet the uncertainty surrounding posthumous personality rights, together with the performance-specific structure and finite duration of sections 38 to 38B, leaves an important gap where deceased performers are concerned.

The international response suggests that the next stage of reform should not be framed simply as an expansion of copyright. It is better understood as a question of digital identity, consent and control. India can preserve the conceptual boundaries of copyright while expressly recognising the performer’s interest in preventing unauthorised synthetic replication. That would bring the law closer to the realities of AI without treating every digital imitation as though it were merely another copyright infringement.

Author: Amrita Pradhan in case of any queries please contact/write back to us via email to content@khuranaandkhurana.com or at  Khurana & Khurana, Advocates and IP Attorney

References

1. Anil Kapoor v. Simply Life India (Delhi High Court, 20 September 2023) CS(COMM) 652/2023, interim order.

2. Arijit Singh v. Codible Ventures LLP 2024 SCC OnLine Bom 2445 (Bombay High Court, 26 July 2024).

3. Asha Bhosle v. Mayk Inc 2025 SCC OnLine Bom 3485 (Bombay High Court, 29 September 2025).

4. Krishna Kishore Singh v. Sarla A Saraogi (Delhi High Court, 11 July 2023) CS(COMM) 187/2021.

5. US Copyright Office, Copyright and Artificial Intelligence, Part 1: Digital Replicas (31 July 2024) https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-1-Digital-Replicas-Report.pdf

6. Nurture Originals, Foster Art, and Keep Entertainment Safe Act of 2025, S 1367, 119th Cong, 1st sess (introduced 9 April 2025) https://www.govinfo.gov/content/pkg/BILLS-119s1367is/pdf/BILLS-119s1367is.pdf; HR 2794, 119th Cong, 1st sess (introduced 9 April 2025) https://www.congress.gov/bill/119th-congress/house-bill/2794

7. Regulation (EU) 2024/1689 of the European Parliament and of the Council (Artificial Intelligence Act), Article 50 and Recital 134 https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=OJ:L_202401689.

8. Ministry of Electronics and Information Technology, Government of India, ‘Government Strengthens Regulatory Framework to Address AI-Generated Deepfakes’ (6 August 2026) https://www.pib.gov.in/PressReleasePage.aspx?PRID=2295500&reg=48&lang=2