Who Owns a Face? Celebrity Personality Rights in the Age of Generative AI

Who Owns a Face? Celebrity Personality Rights in the Age of Generative AI

Evaluating Defamation, the Right of Publicity and IP Protection Under Common Law: An Indian and Comparative Perspective.

 

By Sanjay Kumar Pilla | Topic: Intellectual Property Rights and Technology Law.

Practice Area: Personality Rights, Media & Entertainment Law, Information Technology Law and Tort Law.

PART I: Introduction

Fame has always come with a strange bargain: the more recognisable a person becomes, the less their face seems to belong only to them. For decades this was manageable, because a photograph in a magazine or a poster on a wall was still, in some basic sense, real. Generative artificial intelligence has quietly taken that assumption apart. A synthetic video can now place a celebrity’s face onto another body, a cloned voice can read out lines the actor never spoke, and either can be produced in minutes by someone with a laptop and no filmmaking skill whatsoever. The output does not need to be flattering or even fully convincing on close inspection; it only needs to be convincing enough to be shared before anyone stops to check.

This matters to the law because it collapses problems that used to sit in separate boxes. Passing off a celebrity’s endorsement without consent was a commercial dispute. Publishing a false statement that damaged someone’s reputation was defamation. Copying a photograph or a song without licence was copyright infringement. A single deepfake can now do all three in one stroke, and it can do so faster than a takedown notice can be drafted, let alone acted upon.

What follows is an attempt to trace how Indian courts, working mostly through common law reasoning rather than any dedicated statute, have responded to this problem so far. It also looks at the older common law ideas, largely American, that Indian personality rights jurisprudence has quietly borrowed from, and points out where the gaps still sit.

PART II: The Legal Architecture Behind Personality Rights

2.1 No Standalone Statute, and Why That Has Been Enough So Far

India has no law that uses the phrase “personality rights” or “right of publicity.” What exists instead is closer to a patchwork: the tort of passing off, borrowed from trademark practice; Article 21 of the Constitution, read expansively to cover privacy and dignity; provisions of the Information Technology Act, 2000 aimed at impersonation and obscene content; and ordinary defamation principles drawn from both criminal and civil law. Courts have stitched these together case by case, usually through interim injunctions rather than full trials, because a deepfake that has already gone viral cannot meaningfully be undone by a judgment delivered two years later.

The 2010 Delhi High Court ruling on a doll manufacturer who sold toys resembling the singer Daler Mehndi is where this all really begins. The court held that a well-known individual’s persona carries commercial value in its own right, describing it as a kind of quasi-property interest tied to that person’s identity. At the same time, the judgment drew a boundary: parody and caricature remained protected expression, and only unauthorised commercial exploitation crossed the line.

2.2 The Anil Kapoor Order: Courts Confront Generative AI Directly

The clearest early engagement with synthetic media came in September 2023, when actor Anil Kapoor obtained an ex-parte interim injunction from the Delhi High Court against sixteen named defendants. His complaint described AI tools superimposing his face onto other actors and animated characters, generating merchandise and motivational content in his likeness without consent, and misusing his signature catchphrase for ringtones and GIFs. Justice Prathibha M. Singh granted an in rem order, binding not just the named defendants but the world at large, and directed telecom and IT authorities to block the offending links.

What set this order apart from earlier passing-off cases was its focus on the tools themselves rather than only the finished product. The court restrained the use of AI, face-morphing software, and voice-cloning technology to recreate the actor’s persona, treating the method of generation as part of the wrong rather than a neutral instrument someone happened to misuse. Later personality rights orders in India have largely followed this framing.

The order nevertheless leaves a broader question unanswered. By framing the injunction in expansive terms and extending it against the world at large, the court understandably prioritised preventing immediate harm in an environment where synthetic content can spread within hours. At the same time, the reasoning offers relatively little guidance on where the boundary lies between unlawful commercial exploitation and legitimate expressive uses of AI, such as satire, transformative works or documentary reconstruction. Those questions will almost certainly require fuller consideration in future litigation, particularly where the use of a public figure’s likeness is not primarily commercial but artistic or informational in nature.

2.3 From One Actor to a Pattern

Kapoor’s case did not arrive out of nowhere. Amitabh Bachchan had secured a similar ex-parte order in late 2022 after his name, voice and image turned up without authorisation across various websites and apps. In May 2024, Jackie Shroff won a comparable injunction after an unlicensed AI chatbot began replying to users in a voice modelled on his own, alongside distorted videos and merchandise built around his nickname. Justice Sanjeev Narula again ordered infringing links blocked and again treated the chatbot’s unauthorised use of Shroff’s persona as falling squarely within personality rights, while noting separately that genuine artistic or transformative work still deserved room to exist.

Read together, these orders show a doctrine being built one case at a time rather than waiting on Parliament. A chatbot, a face-swapped video, a cloned voice selling a product the celebrity never touched — each new fact pattern stretches the reasoning from the Daler Mehndi doll case a little further, into technology that did not exist when that judgment was written.

2.4 Defamation and Publicity Are Not the Same Claim

A single deepfake can trigger both wrongs at once, but the law still separates them. A right of publicity claim asks whether someone’s identity was commercially exploited without consent, and it does not matter whether the underlying content is true, flattering, or embarrassing. A defamation claim asks something different: whether a false statement was published that damaged the person’s reputation in the eyes of others. A deepfake showing a celebrity endorsing a product they have never used is a publicity problem even if no viewer thinks worse of them for it. A deepfake putting offensive or criminal words in their mouth is a defamation problem even if no money changes hands anywhere in the chain. Cases involving pornographic manipulation of an actor’s image, which unfortunately make up a sizeable share of the litigation so far, tend to sit at the overlap of both wrongs, with the obscenity and impersonation provisions of the IT Act layered on top.

PART III: The Common Law Roots — Looking Beyond India

Indian courts did not invent this reasoning from scratch. The idea that a person’s identity carries commercially protectable value traces back to American common law, where the right of publicity grew out of privacy tort doctrine over the twentieth century. Two decisions in particular keep resurfacing in Indian judgments and legal commentary.

3.1 Midler v. Ford Motor Co. (1988)

When Ford’s advertising agency could not obtain a licence to use Bette Midler’s own recording for a commercial, it hired a former backup singer to imitate her voice so closely that Midler’s own friends assumed she had performed it herself. The Ninth Circuit held that deliberately copying a distinctive, identifiable voice for commercial purposes, after the artist had refused to lend her actual voice, was an actionable wrong on its own, separate from any copyright claim over the recording itself. The judgment is often cited today as the nearest pre-AI equivalent to voice cloning, since it treats imitation, not just direct reproduction, as capable of violating an identity interest.

3.2 White v. Samsung Electronics America, Inc. (1992)

A Samsung advertisement showed a robot in a wig, gown, and jewellery, posed beside a game board in a way that unmistakably evoked television personality Vanna White, without ever using her name, photograph, or voice. The Ninth Circuit held that a right of publicity claim could succeed even where identity was evoked only indirectly, so long as an ordinary viewer would recognise who was being referenced. Judge Alex Kozinski’s dissent in that case is cited almost as often as the majority opinion; he warned that stretching the publicity right too far could end up chilling parody, commentary, and other legitimate reuse of a public figure’s image. That same tension — protecting identity without swallowing free expression — is what Indian courts have tried to manage when they carve out room for parody even while shutting down commercial deepfakes.

English common law took a more cautious route, never recognising a freestanding right of publicity and instead routing similar claims through passing off, malicious falsehood, or breach of confidence. Indian courts have effectively built something in between: keeping the passing-off framework inherited from England, borrowing the identity-based reasoning of the American cases, and adding constitutional privacy protection on top of both.

PART IV: Why Generative AI Breaks the Old Assumptions

4.1 There Is No Longer Anything to Copy

Older passing-off cases assumed a defendant had used an actual, identifiable photograph, recording, or name belonging to someone else. Generative AI removes that assumption entirely. A diffusion model can output a synthetic photograph of a celebrity that was never captured by any camera and does not correspond to any single training image a court could point to — and the result can still be unmistakably that person. The wrong, in other words, sits in generating a convincing likeness rather than in reproducing an existing one, which is why recent Indian injunctions are written around the use of AI tools to recreate a persona, rather than around any specific photograph or clip.

4.2 Speed Has Outrun the Old Remedies

A defamatory newspaper article could once be answered with a printed correction a week later, and the damage, while real, stayed within some rough bounds. A defamatory deepfake on a messaging platform can reach millions of people within hours, and by the time a takedown notice reaches the platform, the same clip has usually resurfaced under a different filename somewhere else. This gap is exactly why Indian courts have leaned so heavily on dynamic, in rem injunctions that bind unnamed future infringers and instruct internet and telecom intermediaries to block content proactively, instead of relying on the older model of identifying and suing one defendant after the harm is already done.

4.3 Parody Is Getting Harder to Tell Apart from the Real Thing

Courts have generally protected parody and satire, on the reasoning that mimicking a public figure for commentary has never required their consent. Generative AI complicates that protection because a synthetic video can be photorealistic rather than obviously exaggerated, which removes the visual cue that once told an audience they were watching commentary rather than reality. A caricature drawing announces itself as art the moment you look at it; a convincing deepfake, by design, does the opposite.

4.4 The Limited Role of the Digital Personal Data Protection Act, 2023

One statute that might appear relevant at first glance is the Digital Personal Data Protection Act, 2023. The Act regulates the processing of digital personal data and strengthens an individual’s control over information relating to them, including obligations concerning consent and lawful processing. Yet its usefulness in deepfake disputes remains limited. Personality rights concern the commercial and reputational value attached to a person’s identity, whereas the DPDP Act is primarily concerned with protecting personal data and ensuring responsible data processing. A synthetic image or AI-generated voice may not always involve the direct processing of personal data in the manner contemplated by the legislation, particularly where the output is generated from publicly available material or machine-learned patterns rather than by reproducing an existing photograph or recording. As a result, while the Act may complement privacy-based claims in certain situations, it cannot presently substitute for the common law framework governing personality rights and publicity.

 

PART V: Conclusion

None of this adds up to a codified right of publicity in India — what exists instead is a set of interim orders, built one injunction at a time, borrowing structure from a decade-old doll case and vocabulary from American privacy torts written long before deepfakes were a possibility anyone had to plan for. The approach works for the narrow set of people who can afford to reach the Delhi High Court quickly: Kapoor, Bachchan, and Shroff all obtained relief within weeks of filing. It offers considerably less to the person without a recognisable name, whose stolen face or cloned voice may never generate the same urgency in a courtroom.

That gap is really the open question going forward — not whether courts can act fast when a famous plaintiff walks in, since they clearly can, but whether the same protection will ever extend to someone whose identity has no market value to point to, only a private life that has been misused.

Whether that development should continue through judicial innovation alone is open to debate. Indian courts have responded with considerable flexibility, but litigation remains reactive and fact-specific, leaving important questions to be answered only after harm has already occurred. A narrowly tailored statutory framework recognising personality rights, while expressly preserving legitimate journalism, parody, criticism and other forms of protected expression, would provide greater certainty for creators, intermediaries and rights holders alike. Such legislation need not replace the existing common law principles; rather, it could codify the protections that have gradually emerged through judicial decisions while offering clearer standards for AI-generated content in an increasingly digital public sphere.

Statutes & Legal Provisions

• The Constitution of India, Article 21 (right to life, privacy and dignity).
• The Trade Marks Act, 1999, Section 2(m) and Section 29 (as extended to personality indicia).
• The Information Technology Act, 2000, and the Intermediary Guidelines and Digital Media Ethics Code Rules, 2021.
• The Indian Penal Code / Bharatiya Nyaya Sanhita provisions on defamation and impersonation.

Case Law

• M. Entertainment Pvt. Ltd. v. Baby Gift House & Ors., CS(OS) No. 893/2002, Delhi High Court, decided 2010.
• Amitabh Bachchan v. Rajat Nagi & Ors., CS(COMM) 819/2022, Delhi High Court.
• Anil Kapoor v. Simply Life India & Ors., CS(COMM) 652/2023, Delhi High Court.
• Jaikishan Kakubhai Saraf alias Jackie Shroff v. The Peppy Store & Ors., CS(COMM) 389/2024, Delhi High Court.
• Midler v. Ford Motor Co., 849 F.2d 460 (9th Cir. 1988), United States Court of Appeals for the Ninth Circuit.
• White v. Samsung Electronics America, Inc., 971 F.2d 1395 (9th Cir. 1992), United States Court of Appeals for the Ninth Circuit.

Secondary Sources

• LiveLaw, reportage on the Delhi High Court orders in the Anil Kapoor and Jackie Shroff personality rights suits, 2023-2024.
• Jennifer E. Rothman, The Right of Publicity: Privacy Reimagined for a Public World (Harvard University Press, 2018).
• Thomas McCarthy, The Rights of Publicity and Privacy.
• Mark P. McKenna, “The Right of Publicity and Autonomous Self-Definition,” University of Pittsburgh Law Review (2006).
• Intellepedia (BananaIP), case summary of D.M. Entertainment Pvt. Ltd. v. Baby Gift House & Ors.

 

 

Here are some of the most relevant FAQs.

1. Why doesn’t India have a separate law for personality rights?

Answer:
India has developed personality rights through judicial interpretation rather than legislation. Courts have relied on passing off, Article 21 of the Constitution, privacy principles, defamation law, and intellectual property concepts to protect a person’s identity. While this case-by-case approach has been effective in many celebrity disputes, it lacks the certainty of a dedicated statute.

2. Can a celebrity sue someone merely for creating an AI-generated image of them?

Answer:
Not necessarily. The legality depends on the purpose and effect of the image. If it is used for commercial gain, falsely suggests endorsement, damages reputation, or infringes personality rights, legal action may be justified. However, genuine parody, satire, artistic expression, or commentary may still receive constitutional protection under freedom of speech.

3. How are deepfakes different from traditional copyright infringement?

Answer:
Traditional copyright protects original works such as photographs, videos, or recordings. Deepfakes often create entirely new synthetic content without directly copying an existing work. The legal issue therefore shifts from copyright infringement to misuse of a person’s identity, reputation, and commercial value through personality rights and defamation.

4. Why is the Anil Kapoor judgment considered significant?

Answer:
The Delhi High Court was among the first Indian courts to directly address AI-generated misuse of a celebrity’s identity. The judgment recognised that face-swapping, voice cloning, and similar AI technologies could violate personality rights and granted broad injunctive relief against their unauthorized use.

5. Does the Digital Personal Data Protection Act, 2023 solve the deepfake problem?

Answer:
No. The DPDP Act primarily regulates the processing of personal data and consent. While it may support privacy claims in certain situations, it does not specifically recognise personality rights or regulate AI-generated synthetic media. Deepfake disputes therefore continue to depend largely on common law remedies.

6. Are personality rights only available to celebrities?

Answer:
Every individual possesses privacy and identity interests. However, celebrities generally receive stronger protection under personality rights because their identity has independent commercial value. Ordinary individuals are more likely to rely on privacy, defamation, impersonation, or data protection laws rather than publicity rights.

7. Can AI companies be held liable for deepfakes created by users?

Answer:
Potentially, yes. Liability depends on the platform’s role, knowledge of the infringement, compliance with intermediary obligations, and whether it acted promptly after receiving notice. Courts are increasingly examining platform responsibility alongside the conduct of individual users.

8. Should India enact a separate Personality Rights Act?

Answer:
A dedicated statute would provide greater legal certainty by defining personality rights, available remedies, permissible exceptions, and platform obligations. It could also establish safeguards for freedom of expression, journalism, education, and parody while addressing AI-generated misuse more comprehensively.

9. How do Indian courts balance personality rights with freedom of speech?

Answer:
Indian courts generally distinguish between commercial exploitation and legitimate expression. Unauthorized endorsements, deceptive advertisements, and misleading commercial uses are more likely to attract liability, whereas parody, criticism, news reporting, and transformative artistic works may receive constitutional protection, provided they do not unfairly exploit a person’s identity.

10. What is the biggest legal challenge posed by generative AI?

Answer:
The greatest challenge is that existing legal doctrines were developed for real photographs, recordings, and human-created content. Generative AI can produce convincing synthetic images and voices that are not direct copies of any original work, making it difficult to determine ownership, liability, and appropriate remedies under traditional legal frameworks.