INTRODUCTION
Personality rights empower individuals to regulate the usage of their distinctive identity markers such as their name, voice, physical likeness, signature, and unique personal attributes. Identity in the modern world is an assent which has commercial value. With the use of digital media, celebrities can utilize their personalities or identity markers to generate considerable revenue through advertising, brand sponsorships, endorsements, image licensing and social media outreach. In India therefore, the current lack of a dedicated statutory framework that protects personality rights results in the risk of unauthorized exploitation, dilution of reputation, and economic loss.
To add to this challenge, the speed of Generative AI and deepfake-driven identity misappropriation has vastly outpaced the uneven legislative remedies available at present. Currently, deepfakes are cheap and easy to mass-produce and cloning the voice of a well-known singer or using them for endorsing a particular product takes only a few minutes and largely available audio samples. Moreover, such synthetically made endorsement can reach thousands of viewers even before any legal process can be initiated, thereby resulting in widespread commercial exploitation of a celebrity’s personality. There is thus a lack of a unified legal framework for courts nationwide to combat the exploitation of personality rights. The current method operated through interim injunctions and dynamic takedown orders against infringers but with generative AI escalating the scale of infringement the problem has grown to an extent that legislative action can no longer ignore.
Ultimately, a pressing need has arisen for a formalized personality rights statute and definition. This need not only stems from the right of celebrities to protect their commercial interests but also the right of privacy which is a fundamental right available to each and every citizen of the nation.
THE PRESENT REGULATORY FRAMEWORK
To protect the personality rights of citizens, India rather than relying on a single statute, hinges its enforcement mechanism on pre-existing statutory frameworks. As of 2026, High Courts are systematically bridging the legal divide by drawing from a blend of fundamental constitutional rights and statutory protections across the Information Technology, Trade Marks, and Copyright Acts. The following section examines the existing statutory and constitutional mechanisms leveraged by Indian courts to enforce personality rights.
Constitutional Basis:
The right to life and personal liberty guaranteed under Article 21 of the Constitution of India[1] provides the broadest foundation for the enforcement of personality rights. After the Justice K.S Puttaswamy judgement[2], Article 21 has been interpreted by the Apex Court to include the right to privacy as a fundamental right. Emerging from this, the right to dignity, and, by extension, the right to control the exploitation of one’s identity is considered to be a large part of personality rights jurisprudence. In contrast, Article 19(1)(a) guarantees freedom of speech and expression, creating a tension with personality rights claims that courts must balance with satire, parody, commentary, and journalism.
Copyright Act:
Under Section 17 of the Copyright Act,[3] the author of an original literary, dramatic, musical or artistic work is recognized as the original first owner of copyright. When Generative AI models scrape copyrighted portraits, photographs, audio tracks, or videos to produce deepfakes, they also infringe on the underlying copyright of the photographer or creator or the subject who had retained ownership. However, Section 17(c) of the said Act also dictates that if a work is created during the course of employment or under a contract of service, the employer is the first owner. Hence actors and models, frequently do not hold copyright in their own recorded likenesses, as it would technically belong to the film production house or studio. Thus, they are compelled to look beyond Section 17 towards performer and moral rights.
The Copyright Act does not define the word ‘personalities’ or ‘celebrities’. However, it provides the definition for the term ‘performer’ under Section 2(qq) as “actor, singer, musician, dancer, acrobat, juggler, conjurer, snake charmer, a person delivering a lecture or any other person who makes a performance.” A celebrity can aptly fall within the ambit of a ‘performer’ and thus be entitled to rights construed to a ‘performer’. Section 38 of the Copyright Act defines ‘Performer’s right’ and states that performer’s right shall be available for a period of 50 years from beginning of calendar year following the year in which the such performance was made. Moreover, Section 38A gives performers the exclusive right to make sound or visual recordings of their live performance and to reproduce, issue copies of, or communicate that performance to the public. When an AI voice-cloning tool or diffusion model is trained on an artist’s sound recordings to generate new, synthetic tracks, as seen in Arijit Singh v. Codible Ventures[4], the unlicensed extraction and re-synthesis directly violate the performer’s exclusive reproduction and public communication rights under Section 38A. Section 39 which deals with acts not infringing performer/broadcaster’s rights further defines specific statutory carve-outs where performer rights are not infringed, including private or personal use, reporting current events, or bona fide teaching and research none of which justify Generative AI exploitation.
Furthermore, Section 52 which is the fair dealing provision limits copyright infringement claims when the works are used for private or personal research, criticism, review or reporting of current events and for parody, satire and caricature. Developers often attempt to shield large-scale AI data scraping under this fair dealing provision through the guise of research or text-and-data mining. However, Section 52 of the Indian Copyright Act is an exhaustive list of specific exceptions, unlike the open-ended Fair Use doctrine of the United States.[5] Indian courts, through the plethora of judicial decisions discussed in the next section, have consistently held that commercial AI model training and unauthorized commercial personality synthesis fall well outside the narrow scope of Section 52 fair dealing.
Apart from the performer’s right route, the moral right framework in Sections 38B and Section 57 also offer a strong anchor for personality rights protection. The Right to Paternity grants creators and performers the right to claim authorship credit and restrain false attribution and thereby prevents infringers from publishing AI-generated audio or video under an artists name without consent. The Right to Integrity further grants authors the absolute right to claim damages for any distortion or other derogatory acts to their works. This is best demonstrated in the case of Amar Nath Sehgal v. Union of India[6] which established that moral rights survive even if the economic right surrounding the same work has been transferred. This framework can be used to guarantee that even if an artist does not own their original work then also their reputation must remain protected.
[3] The Copyright Act, 1957 (Act 14 of 1957), s.17.
[4] Arijit Singh v. Codible Ventures LLP and Ors., Interim Application (L) No. 23548 of 2024 in Commercial IP Suit (L) No. 23443 of 2024, decided on 26-07-2024 (Bom HC).
[5] 17 U.S. Code § 107.
[6] Amar Nath Sehgal v. Union of India and Anr., 2005 (30) PTC 253 (Del).
Trademarks Act:
Beyond copyright, the Trademarks Act, 1999[7] is another statutory mechanism which can be utilized for the protection of personality rights. As per Section 2(1)(m) of the Act, the definition of the word ‘mark’ is inclusive of names, which grants individuals, particularly celebrities and public figures, the locus standi to register their names, signatures, or distinctive monikers as commercial brand identifiers. Furthermore, to prevent the commercial exploitation of said brand identifiers, Section 14 of the Act directs that any trademark application seeking to register or commercially exploit the name or representation of a living person, or a person who died within twenty years prior to the application, requires prior written consent from that individual or their legal heirs. To strike a balance between protection of personality rights and the freedom of expression, Section 35 of the Act guarantees that no registered trademark owner can interfere with the bona fide use by any person of their own name or the name of their place of business.
[7] The Trade Marks Act, 1999 (Act 47 of 1999).
Code of Self-Regulation in Advertising in India
Chapter IV of the Code of Self-Regulation in Advertising in India[8] released by the Advertising and Standards Council of India (ASCI) states the following:
“4.2 Advertisements shall not make unjustifiable use of the name or initials of any other firm, company or institution, nor take unfair advantage of the goodwill attached to the trademark or symbol of another firm or its product or the goodwill acquired by its advertising campaign.”
This provision be construed to ensure that no AI generated content can be created to endorse a particular product which a celebrity has not agreed by misrepresenting the goodwill attached to their name and reputation.
[8] Advertising Standards Council of India, The Code for Self-Regulation of Advertising Content in India, ch. IV.
The Information Technology Act, 2000 and the corresponding IT Rules, 2021
The Information Technology Act, 2000[9] (IT Act, 2000) introduces civil and criminal sanctions for digital impersonation and privacy violations in Sections 66C, 66D and 66E, respectively, in particular, for identity theft, cheating by impersonation and privacy violations. By interpreting a ‘unique identification feature’ mentioned in Section 66C of the IT Act, 2000, broadly to include facial geometry and voice biometrics, Indian courts have brought AI-generated deepfakes and voice clones within the scope of Section 66C violations.
Furthermore, the 2026 amendment to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021[10] (IT Rules, 2021) introduced the Synthetically Generated Information (SGI) framework as the government’s first direct legislative measure targeting AI generated content. SGI covers any audio, visual, or audiovisual content artificially or algorithmically created or altered to appear indistinguishable from a natural person or real-world event[11]. Under this regulatory framework, platforms permitting the creation or dissemination of SGI are obligated to mark such content with explicit text and audio disclosures, as well as embed unique cryptographic identifiers traceable to the originating device.[12]
[9] The Information Technology Act, 2000 (Act 21 of 2000).
[10] The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026, G.S.R. 120(E).
[11] The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, r. 2(1) (wa) (as inserted by the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026, r. 2,).
[12] The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, r. 3(3)(a)(ii) (as inserted by the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026, G.S.R. 120(E)).
EVOLUTION OF JUDICIAL APPROACH TO PERSONALITY RIGHTS IN THE AGE OF GENERATIVE AI
The very first case which dealt with constitutional privacy jurisprudence in terms of personality rights is the Supreme Court’s ruling in R. Rajagopal v. State of Tamil Nadu (1994)[13], in which personal identity was tied intricately with Article 21. Since then, the protection of a personality was largely rooted in the common-law tort of passing off and privacy. However, the rise of digital platforms and generative AI triggered has a shift towards a more proactive, injunctive relief against the unauthorized commercial and synthetic exploitation of public personas.
Personality rights first came into focus in the case of Amitabh Bachchan v. Rajat Negi & Ors. (2022)[14], where the Delhi High Court granted a first of its kind John Doe injunction against all infringing websites and intermediaries. The Court protected the actor’s voice, image, and persona from unauthorized exploitation across both traditional commerce and emerging technologies like NFTs and the Metaverse. Following this case, in Anil Kapoor v. Simply Life India & Ors. (2023),[15] the Delhi High Court explicitly brought synthetic AI Generated media into focus, ruling that unauthorized AI deepfakes, synthetic voice clones, and the exploitation of iconic catchphrases infringe upon a celebrity’s personality rights and their fundamental right to livelihood under Article 21.
During 2024, the courts refined the boundaries between actionable commercial misappropriation and creative expression. In Jaikishan Kakubhai Saraf (Jackie Shroff) v. The Peppy Store & Ors. (2024),[16] the Delhi High Court protected Shroff’s likeness, voice, and trademarked slang from commercial merchandising and abusive AI chatbots, affirming that while bona fide parody is protected, commercializing a persona to drive digital traffic crosses into infringement. Concurrently, in Karan Johar v. India Pride Advisory Pvt. Ltd. (2024–2026), [17] the Bombay High Court extended these protections to prominent filmmakers, confirming that directorial identity cannot be used as an unauthorized promotional vehicle.
The jurisprudence rapidly adapted to specialized generative AI models in Arijit Singh v. Codible Ventures LLP (2024).[18] The Bombay High Court restrained platforms from utilizing Real Voice Cloning (RVC) models trained on the artist’s vocal repertoire, establishing that an individual’s vocal timbre is an indispensable facet of identity protected under both personality and copyright moral rights. Similarly, the Delhi High Court extended John Doe protections to fictional personas in Neela Telefilms v. John Doe (Taarak Mehta Ka Ooltah Chashmah Case),[19] restraining the dissemination of AI-generated deepfakes and sexually explicit morphs of fictional characters.
Most recently, in Preity G. Zinta v. Google LLC & Ors. (2026), [20] the Bombay High Court reinforced digital personality protection. Addressing the widespread AI deepfakes and voice-simulating chatbots of Ms. Zinta the Court stated that the unauthorized commercialization of distinctive traits can count has harm to dignity and trigger protection under Article 21.
These decisions seen together, illustrate an evolution from basic common-law tort protections to a robust, constitutionally backed personality rights framework engineered to counter synthetic and algorithmic exploitation.
[13] R. Rajagopal v. State of Tamil Nadu, (1994) 6 SCC 632.
[14] Amitabh Bachchan v. Rajat Nagi & Ors., 2022 SCC OnLine Del 4110.
[15] Anil Kapoor v. Simply Life India & Ors., 2023 SCC OnLine Del 6914.
[16] Jaikishan Kakubhai Saraf alias Jackie Shroff v. The Peppy Store & Ors., 2024 SCC OnLine Del 3975.
[17] Karan Johar v. India Pride Advisory Pvt. Ltd. & Ors., Interim Application (L) No. 17792 of 2024 in Commercial IP Suit (L) No. 17788 of 2024 (Bom HC).
[18] Arijit Singh v. Codible Ventures LLP & Ors., 2024 SCC OnLine Bom 2422.
[19] Neela Film Productions Pvt. Ltd. v. TaarakMehtaKaOoltahChashmah.com & Ors., CS(COMM) 674/2024 (Del HC).
[20] Preity G. Zinta v. Google LLC & Ors., Commercial IP Suit No. 142 of 2026 (Bom HC).
GLOBAL REGULATORY RESPONSES TO AI-DRIVEN PERSONALITY INFRINGEMENT
United States
The US approach to the protection of personality rights is heavily commercially focused, recognizing a common law tortious action against individuals who infringe upon a person’s “Right of Publicity”[21]. However, there is no federal legislation governing personality rights and hence the scope and application of these rights remain inconsistent across different states. The right was first recognized in Haelan Laboratories v. Topps Chewing Gum (1953)[22]. It was expanded in Vanna White v. Samsung Electronics (1993)[23], concerning the TV personality Vanna White from Wheel of Fortune, where courts ruled that a television presenter’s signature characteristics or “get-up” is eligible for protection from unauthorized commercial exploitation. Recently, Generative AI has this brought this issue into the federal spotlight. The State of Tennessee’s 2024 ELVIS Act[24] became the first US law to explicitly extend right of publicity protection to an individual’s voice against AI cloning. Furthermore, the NO FAKES Act,[25] unanimously advanced by the Senate Judiciary Committee in June 2026, would create the first federal intellectual property right in a person’s voice and visual likeness against unauthorized digital replicas, with a dedicated notice-and-takedown mechanism modelled on the DMCA.
[21] Melville B Nimmer, ‘The Right of Publicity’ (1954) 19(2) Law and Contemporary Problems 203
[22] Haelan Laboratories, Inc v Topps Chewing Gum, Inc 202 F 2d 866 (2d Cir 1953).
[23] Vanna White v Samsung Electronics America, Inc 971 F 2d 1395 (9th Cir 1992)
[24] The Ensuring Likeness Voice and Image Security Act of 2024, 2024 Tenn. Pub. Acts, ch. 588 (US).
[25] Nurture Originals, Foster Art, and Keep Entertainment Safe Act of 2026 (NO FAKES Act), S. 4591, 119th Cong. (2026) (US).
United Kingdom
Unlike the US, the United Kingdom has never had a statutory “Right to Publicity” or general personality right. The method of protection usually adapted in the UK is the tort of passing off, which was crystallized in the case of Irvine v Talksport Ltd (2002),[26] where the UK Court of Appeal held that unauthorized use of a celebrity’s image to signal endorsement is harmful to the goodwill established by said celebrity and could lead to reputational damage and misrepresentation. This gave well-known individuals a limited remedy against false endorsement.
The more significant development, however, came through the tort of misuse of private information, which took shape after the Human Rights Act 1998[27] incorporated Article 8 (right to private life) and Article 10 (freedom of expression) of the European Convention on Human Rights (ECHR)[28] into domestic law. Following this came cases such as such as Campbell v MGN Ltd (2004)[29] and Douglas v Hello! Ltd (2007)[30] which established a two-stage test to determine whether the misuse of private information was actually occurring – the first factor is deciding whether the claimant had a reasonable expectation of privacy and the second one is a proportionality balancing exercise against the public’s freedom of expression. This has since become the dominant framework for image-based and personal-information claims, including cases with clear commercial or reputational stakes.
Breach of confidence,[31] image rights merchandising agreements,[32] and, more recently, data protection law under the UK GDPR[33] particularly rights concerning the processing of personal data and images supplement this framework. The result is a framework that protects personality-adjacent interests through privacy, confidence, passing off, and data protection, rather than through a codified legislative right.
This framework has also incorporated provisions to address the consequences of the misuse of generative AI. Section 138 of the Data (Use and Access) Act 2025[34] has made it a criminal offence to create or request the creation of an AI-generated intimate image without consent, further enhancing the protection offered by the Online Safety Act 2023[35] which had criminalized the sharing of such deepfakes by inserting a new section into The Sexual Offences Act, 2003[36].
[26] Irvine v Talksport Ltd [2003] EWCA Civ 423, [2003] 2 All ER 881
[27] The Human Rights Act, 1998.
[28] The Convention for the Protection of Human Rights and Fundamental Freedoms, 1950.
[29] Campbell v MGN Ltd [2004] UKHL 22.
[30] Douglas v Hello! Ltd (No 3) [2007] UKHL 21, [2008] 1 AC 1.
[31] Coco v A N Clark (Engineers) Ltd [1968] FSR 415 (Ch)
[32] Fenty v Arcadia Group Brands Ltd (t/a Topshop) [2015] EWCA Civ 3, [2015] 1 WLR 3291
[33] The UK General Data Protection Regulation, 2018, art. 4(1), art. 9 & art. 17.
[34] The Data (Use and Access) Act, 2025, s.138.
[35] The Online Safety Act, 2023, s. 188.
[36] The Sexual Offences Act, 2003, s. 66B.
Germany
Germany maintains exceptionally strong, comprehensive statutory and constitutional protections for personality rights, which include the integrated right to one’s own image or likeness. These rights are constitutionally guaranteed under Articles 1 and 2 of the German Basic Law[37] and have been protected in Federal Court case law since 1954,[38] notably established in the Marlene Dietrich case.[39] Section 22 of the German Copyright in Works of Fine Art and Photography (KUG)[40] explicitly mandates that consent is required to display or publish any portrait of a person. Section 23 outlines narrow statutory exceptions, such as incidental inclusion in a public setting or the public interest surrounding public figures.[41]
German courts have already applied this framework to generative AI. On 20 August 2025, the Berlin Regional Court held that AI-cloning a voice actor’s distinctive dubbing voice without consent violated his personality right under Article 2(1) read with Article 1(1) of the Basic Law, and further awarded the litigant license-fee damages[42].
[37] The Basic Law for the Federal Republic of Germany, 1949, art. 1 & art. 2.
[38] Corinna Coors, “Celebrity Image Rights Versus Public Interest: Striking the Right Balance under German Law” 9 Journal of Intellectual Property Law and Practice 835 (2014).
[39] 63 BGH, 1 December 1999, I ZR 49/97 Marlene Dietrich—BGHZ 143, 214.
[40] Gesetz betreffend das Urheberrecht an Werken der bildenden Künste und der Photographie (Kunsturhebergesetz, KUG) 1907, s 22 (Germany).
[41] Caroline von Hannover v Germany App no 59320/00 (ECtHR, 24 June 2004).
[42] Landgericht Berlin II (Berlin Regional Court II), judgment of 20 August 2025, Case No. 2 O 202/24.
France
France possesses a robust, privacy-focused image right (droit à l’image) that prevents the reproduction or distribution of a person’s image without consent. This right is derived from the broad privacy protections of Article 9 of the French Civil Code.[43] The French framework focuses heavily on protecting the individual’s dignity and private sphere.[44] For example, French actor Olivier Martinez utilized Article 9 to bring action in the UK against an English newspaper for violating his image rights.[45]
France has since codified a specific criminal response as well. The SREN Law of 21 May 2024[46] amended Article 226-8 of the Penal Code[47] to expressly cover deepfakes and other content generated by algorithmic processing, and introduced a dedicated Article 226-8-1 imposing aggravated penalties, of up to two years imprisonment and a €60,000 fine, for non-consensual sexual deepfakes, with exceptions preserved for content that is obviously satirical, artistic, or fictional.
[43] The French Civil Code, 1804, art. 9.
[44] Elisabeth Logeais and Jean-Baptiste Schroeder, ‘The French Right of Image: An Ambiguous Concept Protecting the Human Persona’ 18 Loyola of Los Angeles Entertainment Law Review 511 (1998).
[45] Joined Cases C-509/09 and C-161/10 eDate Advertising GmbH v X and Olivier Martinez and Robert Martinez v MGN Limited [2011] ECR I10269
[46] The Law to Secure and Regulate the Digital Space (SREN Law), 2024.
[47] The French Penal Code, 1994, art. 226-8.
Belgium
Belgium protects a hybrid personality and publicity right that controls the publication of one’s own image, covering both the physical likeness and aspects of a person’s reputation. Unlike France and Germany, this right is not explicitly in the Constitution but is codified in Article XI.174 of the Belgian Code of Economic Law (BCEL)[48] and strictly mandates prior consent for both the taking and distributing of images. This right mandates that the author of a portrait, the owner of a portrait and any other bearer or holder of a portrait will not be able to reproduce it or communicate it to the public without the consent of the person who has been depicted in the same during the time of their life or for 20 years after the person’s death, without the consent of his heirs. It further allows victims to sue for both damages and immediate image removal.
Belgium has not yet enacted a dedicated national deepfake statute, instead relying on the directly applicable EU AI Act[49] to govern AI-generated likeness. Article 50 of the Act obliges providers to mark synthetic content in a machine-readable format and requires deployers to disclose deepfakes, defined broadly enough to capture any realistic AI-generated depiction of an existing person regardless of deceptive intent.
Guernsey
In 2012, Guernsey became the first jurisdiction to enact a statute specifically dedicated to image rights through the Image Rights Ordinance, 2012.[50] This is a highly commercially focused registration system which allows natural persons, joint personalities, and even fictional characters to register their personalities to protect all associated images.[51] However, since it is commercially driven, infringement only occurs if the registered image is used without consent “for a commercial purpose or economic benefit,”[52] making it less effective for purely non-commercial privacy violation.
[50] The Image Rights (Bailiwick of Guernsey) Ordinance 2012.
[51] ibid, s 1 (definitions of “personnage,” including natural persons, joint personalities, groups, legal persons, and fictional characters).
[52] ibid, s 27(1) (infringement occurs only where a protected image is used for a commercial purpose or a financial or economic benefit).
European Union: The AI Act as a Convergence Point
The European Union’s Artificial Intelligence Act[53] does not create a substantive personality right but rather imposes a procedural obligation through its Article 50. It states that online intermediaries or internet service providers must ensure AI-generated content is marked in a machine-readable, detectable format, and deployers of deepfakes, defined in Article 3(60) as AI-generated content that resembles a real person and would falsely appear authentic,[54] must clearly disclose that fact, irrespective of deceptive intent. This labelling-first approach is instructive for India’s own previously discussed SGI framework under the 2026 IT Rules amendment[55].
Comparative Summary: Personality Rights and AI Across Jurisdictions
Country | Core Legal Basis | Key Judicial Rulings | AI-Specific Development |
India | Constitutional (Art. 21), Copyright Act, Trade Marks Act, IT Act/Rules | Amitabh Bachchan v Rajat Negi (2022); Anil Kapoor v Simply Life India (2023); Arijit Singh v Codible Ventures (2024) | 2026 IT Rules amendment introduces Synthetically Generated Information (SGI) labelling and traceability framework |
United States | State common-law right of publicity. | Haelan Laboratories v Topps Chewing Gum (1953); Vanna White v Samsung Electronics (1993) | Tennessee ELVIS Act (2024); Upcoming NO FAKES Bill advanced by Senate Judiciary Committee (June 2026) |
United Kingdom | Passing off; misuse of private information, Art. 8 and 10 of the ECHR | Irvine v Talksport Ltd (2002); Campbell v MGN Ltd (2004); Douglas v Hello! Ltd (2007) | Data (Use and Access) Act 2025, s 138 criminalises the generation of AI-generated intimate images |
Germany | Constitutional (Arts. 1 & 2 Basic Law); KUG ss 22–23 | Marlene Dietrich (2000) | Berlin Regional Court voice-cloning ruling (Aug 2025)[56] |
France | Droit à l’image under the French Civil Code, art 9 | Martinez v MGN Ltd (2011, CJEU reference) | SREN Law (2024) amends the French Penal Code art 226-8 by inserting new art 226-8-1 for sexual deepfakes |
Belgium | BCEL art XI.174 | Peter C. v RTL-TVI (Brussels Civil Court, 2001)[57] | Governed by EU AI Act, art 50 (transparency/labelling of AI generated content) |
Guernsey | Image Rights (Bailiwick of Guernsey) Ordinance 2012, s 27 | No reported case law to date | No AI-specific amendment to date. |
European Union | No unified personality right | No reported case law to date. | EU AI Act art. 50, mandatory machine-readable marking and deepfake disclosure. |
KEY SUBSTANTIVE GAPS IN AI DRIVEN PERSONALITY RIGHTS JURISPRUDENCE
Thus, we see that courts in India and over the world have stepped in to curb AI deepfakes and likeness theft, relying heavily on case-by-case injunctions and a developing legislative framework. However, this has left major blind spots. Without a particular statutory base, the legal landscape remains unpredictable. Moving forward, the law must confront some key structural gaps which include the following:
- First of all is the attribute question. As of now, the protection against commercial exploitation has been extended to the name, image, voice, vocal style, mannerisms, catchphrases, and AI-generated replicas of established celebrities. However, the scope of protection for ordinary citizens without commercial recognition remains undefined. The principal shortcoming is that protection depends substantially on whether the victim has acquired sufficient commercial recognition, ordinary citizens without celebrity status are thus left without a suitable remedy in case their personality right is violated.
- Second is the post-mortem rights question. Personality rights in India have no defined duration after death. The Copyright Act provides moral rights protection for the author’s lifetime, but there is no comparable post-mortem personality rights framework. For deceased performers, particularly those whose voice or likeness is commercially valuable to AI training companies, the legal position is uncertain.
- Third is deciding how to balance the freedom of expression and the protection of personality rights. The courts interpret the balance between personality rights and Article 19(1)(a) freedom of expression on an ad hoc basis. A clear definition of permitted exceptions to protection, such as satire, parody and news commentary, and a clear definition of exceptions in academic contexts, would create a more suitable framework for reducing uncertainty for creators who wish simply use celebrity personalities for bona fide reasons and not commercially exploit the same.
- Fourth, the AI training consent framework. The most commercially significant question is whether using a public figure’s voice recordings, video performances, or published images to train an AI model constitutes an actionable use of personality rights or performer’s rights without consent. Current Indian law provides no clear answer. A licensing regime, similar to what collecting societies provide for music performance rights, would enable commercial AI development while protecting the individuals whose data is being used. There is also a need for a robust reporting mechanism to effectively deal with complaints in a time bound manner.
- Fifth, stakeholders also need address the liability gap to answer the question of whether the liability lies with the developer who trained the AI model, the platform hosting the deepfake (Meta, X, YouTube), or the user prompt-engineer?
RECOMMENDATIONS AND SUGGESTIONS
Indian courts have built personality rights almost entirely through commercial misappropriation reasoning as seen in the Amitabh Bachchan, Anil Kapoor and Jackie Shroff case. However, this right is rooted in economic or commercial harm be-falling a recognizable and monetizable personality built through fame. As a result, individuals who are non-celebrities are excluded from this range of protection because they have no “brand equity” to dilute. But the harm to an ordinary citizen from a deepfake which could include harassment, sexual exploitation and reputational damage is often more acute since they lack the resources or platform leverage to get a 72-hour takedown.
A possible recommendation for this issue would to be to separate personality right from the concept of commercial recognition. It would be better to either develop a privacy based “image-rights” frame-work which is not dependent on passing-off or economic-loss. This right would create a statutory form of protection including a consent for likeness use and a right to object which applies regardless of fame.
To answer the post-mortem rights question, it would be better to a define a post-mortem term rather than leave it undefined. Belgium’s 20-year post-mortem term is a workable model based on which India could adopt a similar fixed term. Enforcement of the same would vest in legal heirs, mirroring Section 57 of the Indian Copyright Act.
A statutory framework with explicit carve-outs for satire, parody, news commentary, and academic use could be imported straight from Section 52 of the Indian Copyright Act’s Fair Dealing Section. This closed list of carve-outs would ensure that courts can apply a predictable check-list instead of relying on proportionality analysis on a case-by-case basis.
To address the AI training consent framework a compulsory or voluntary collecting-society licensing regime would be the best possible option. The same could be modelled on Section 33 of the Copyright Act which deals with copyright societies. This would be more efficient than building a new regulatory framework from the ground up. Since India already has statutory machinery for collective licensing of performer or broadcaster rights the same could be extended to AI-training use. It would, however, need to be paired with a mandatory disclosure obligation on developers enforced through the same SGI labelling infrastructure the 2026 IT Rules amendment already created.
To address the liability question we need to approach the same from a three-pronged approach of culpability.
A developer’s liability arises where the model itself was trained without consent on protected likeness data such as the Section 38A infringement in Arijit Singh v. Codible,[58] where the developer is liable for the unauthorized creation of the tool, independent of how it is later used. Platform liability is appropriate where the harm is dissemination, not creation. India’s SGI framework already assigns platforms a labelling and traceability duty and furthermore the EU AI Act Article 50 does the same. Lastly, user or prompt-engineer liability arises where the harm is in the specific output for example if a user prompts a general-purpose model to generate a non-consensual sexual deepfake of a real person. The UK’s Section 138 Data Use and Access[59] offence and France’s Article 226-8-1[60] both target this layer directly through the person who creates or requests creation of the specific harmful output, regardless of who built the underlying model.
CONCLUSION
To conclude, the above discussion shows us that generative AI has moved faster than the law meant to contain it. Courts around the world have borrowed from copyright, trademark, and constitutional privacy rights to grant injunctive relief on an ad hoc basis for personality rights misuse. This has worked reasonably well for celebrities with the resources and brand value to pursue litigation, but it says very little about what happens to an ordinary person whose likeness is misused. There also arise further queries around post-mortem rights, AI training consent, and the exact lability of a deepfake which causes reputational harm which need further clarity. It is thus a need of the hour for policymakers to act. India does not need to build a framework from scratch but simply to draw from an already present foundation. The technology will not wait, and neither should the law, so the time to build this framework is now.
Authored by:
Sampreeti Nath
Master of Laws, Intellectual Property Law, National Law University and Judicial Academy, Assam
GLIPA india’s AI & Emerging Technologies Committee
