India's personality-rights orders against AI misuse, 2022-2026: an overview

Personality rights, or the right of publicity, protect a person's name, image, voice, likeness and other identifying traits from commercial exploitation or demeaning use. In India they are not defined in any statute. They have grown out of judgments of High Courts, mostly the Delhi High Court, which in Titan Industries Ltd. v. Ramkumar Jewellers (2012) recognised the publicity rights of a celebrity, and that decision was the precedent relied on in the Amitabh Bachchan order of 25 November 2022.² That order, passed ex parte against websites, merchants and unknown persons who used the actor's name, voice and image to promote goods and fake lotteries, became the template for what the LiveLaw author Saurabh Gupta calls the "celebrity scramble" for personality rights.¹¹

AI changed the problem. Cheap voice-cloning, face-swapping and chatbot services made it easy to produce convincing audio, video and conversation in a real person's style at scale. The Delhi High Court's order for Anil Kapoor on 20 September 2023 is the earliest verified here to name these tools. It restrained defendants from misusing his persona "using technological tools such as Artificial Intelligence, Machine Learning, deep fakes, face morphing, GIFs", and relied on privacy, passing off and dilution.¹ Since then the same formula has been repeated in orders for many other public figures, and the courts have moved from images and merchandise to voices, chatbots and fabricated endorsements.

The table lists the orders verified for this wiki. It is a selection and not a complete record, because many more celebrities have obtained similar orders. Entries for several of the matters appear separately (Amitabh Bachchan, Anil Kapoor, Jackie Shroff, Arijit Singh, Sadhguru, the Bachchan family, Asha Bhosle, Sonakshi Sinha and Dr Devi Prasad Shetty).

The Delhi High Court accounts for most of the orders, and its 2026 orders for Swami Ramdev, Kajol, Sonakshi Sinha, Varun Dhawan and Ravi Kishan were all passed by Justice Jyoti Singh. The Bombay High Court produced the voice-cloning orders for Arijit Singh and Asha Bhosle and then, in October 2025, two orders by Justice Arif S. Doctor covering AI-generated content, deepfakes and voice-cloned audio.⁴⁵

Despite different judges, the orders now follow a standard pattern.

• Ex parte, against John Doe. Nearly every order is passed without hearing the defendants. They name identified sellers, platforms and accounts and add unknown persons ("Ashok Kumar") so the injunction can reach new infringers. Pre-institution mediation is waived under Section 12A of the Commercial Courts Act on the ground of urgent interim relief, citing Yamini Manohar v. T.K.D. Krithi.³⁷
• AI is named. Restraints typically cover use of a person's name, image, voice and likeness "through any technology, including AI, generative AI, machine learning, deepfakes and face morphing". The Bombay orders add "voice cloned audio" and metaverse environments.⁴⁵
• Platforms take down within hours and disclose identities. Deadlines have shortened from a week in the Bombay orders to 72 hours in the Aishwarya Rai Bachchan and Kajol orders and 36 hours in the Warikoo and Varun Dhawan orders. Google, Meta, X, Amazon and domain registrars must also file basic subscriber information (names, emails, IP logs, payment and registration details), often in a sealed cover.³⁸⁹
• Dynamic relief. Orders cover "similar or identical" content and let the plaintiff notify new URLs for removal without a new suit. The Sadhguru order (see its own entry) calls this a "dynamic+" injunction, with a 36-hour window for new deepfakes, and the Kumar Sanu order requires Google and Meta to act on mirror content within 48 hours or state objections.⁴⁶
• The state is a defendant. MeitY and the Department of Telecommunications are impleaded as proforma defendants and told to issue blocking directions to telecom and internet service providers, which gives the orders reach beyond the private parties.⁸⁹
• Platforms are released after complying. In the Aishwarya Rai Bachchan and Abhishek Bachchan suits, Etsy, Janitor AI and TeePublic were deleted as defendants in May 2026 after filing compliance affidavits and undertaking to follow future takedown orders.

Because there is no statute, the orders pile several causes of action on top of one another.

Personality and publicity rights. The core is a judge-made right to control the commercial use of one's identity. The Anil Kapoor order treats dilution, tarnishment and blurring as actionable torts and says the court "cannot turn a blind eye" to misuse of a personality's name.¹ The Aishwarya Rai Bachchan order defines it as the right to control and protect the exploitation of one's attributes, with a commercial facet and a dignity facet.

Passing off. Plaintiffs plead that fake endorsements, official-looking sites and merchandise mislead the public into thinking the celebrity approved them. This ground is strongest against merchandise and scam sites, and weaker for fan content or parody.³⁸

Article 21 (privacy and dignity). Following R. Rajagopal, the Anil Kapoor order draws on the right to privacy in Article 21, and the Kajol and Swami Ramdev orders cite Articles 19(1)(a) and 21. The Abhishek Bachchan order finds that depicting him in misleading, derogatory or inappropriate settings intrudes on his privacy and dignity. This ground matters most for pornographic or demeaning deepfakes, where there is no commercial exploitation to speak of.⁷⁸

Copyright Act, 1957 (performers' and moral rights). Plaintiffs invoke the performer's rights and moral rights in Sections 38, 38A and 38B. The Kajol order lists these sections, and the Arijit Singh and Jackie Shroff suits pleaded performers' or moral rights as well, though the Arijit Singh order does not decide the moral-rights claim.⁸

Trade Marks Act, 1999. Where the plaintiff has registered a name or sobriquet as a mark, plaintiffs also plead infringement and passing off under Sections 28 and 29. The Kajol order refers to her KAJOL registrations, the Warikoo order to the registered "Warikoo" mark in Classes 35 and 41, and the final decree for Dr Devi Prasad Shetty rests partly on infringement of Narayana Health's marks under Section 29.³⁸

Free speech as a limit. Courts acknowledge Article 19(1)(a). The Anil Kapoor order accepts satire and criticism up to a point, the Jackie Shroff order declined to enjoin a YouTuber's compilation at the ex parte stage, and the Swami Ramdev order deferred takedown of URLs that intermediaries said were parody or satire.¹⁷

Three types of AI harm recur in the orders.

1. Voice cloning. The Arijit Singh, Asha Bhosle, Kumar Sanu and Akshay Kumar orders restrain services that train on a singer's recordings or convert any voice into the celebrity's. The Bombay High Court has held that making such tools available without consent appears to violate personality rights, and the Kumar Sanu order restrains Jammable Ltd, a voice-conversion service of the same name as the jammable.com site named in the Arijit Singh suit, along with other generative-AI voice services.⁴⁶
2. Deepfakes in fraud and false endorsement. The Warikoo, Sadhguru, Swami Ramdev and Dr Devi Prasad Shetty cases involve fabricated videos used to sell investments, products or medicines to the public. These cases mix personality rights with consumer-protection concerns, and in the Warikoo case victims reportedly lost money.³⁷
3. Chatbots and AI companions. The Aishwarya Rai Bachchan, Kajol and Sonakshi Sinha orders target chatbots on platforms such as Janitor AI, spicychat.ai, talkie-ai.com and Character.AI that present as, or speak in the voice of, a real woman, sometimes with obscene or sexual content. These orders treat the platform operator as a defendant that must take down specific chatbots, without deciding whether it is an intermediary entitled to safe harbour.⁸

None of the orders decides whether training a model on a person's images or recordings is itself infringing, which is the question in the news-agency copyright suit ANI v. OpenAI. The orders restrain outputs and services that imitate a person, and they leave the training-data question to copyright law.

Commentators have raised four concerns.

• No statute. In an October 2025 LiveLaw analysis, advocate Saurabh Gupta argued that personality rights should be defined by Parliament, not by piecemeal orders, and that Indian law mixes a privacy-based theory (compensation for dignity harm) with a property-based theory (recovery of the infringer's profits), without choosing between them.¹¹
• Overbreadth and chilling effect. The same author warns that sweeping ex parte John Doe orders can restrain satire, parody, criticism and news reporting, and recommends express carve-outs, a higher evidentiary threshold for urgent relief and narrower restraints.¹¹
• Recognisability as infringement. A June 2025 SpicyIP critique of the Sadhguru and Warikoo orders says the courts presumed infringement from recognisability alone, without testing confusion or harm, and that a 36-hour compelled takedown through an ex parte process risks over-removal because satire and fair use are never argued. It suggests a confusion test, an academic amicus before blanket ex parte orders, and a clear distinction between privacy-based and publicity-based claims.¹²
• Voice cloning and open questions. SpicyIP's commentary on the Arijit Singh order (see its own entry) asked how courts will treat parody and remixes, what standard protects non-famous people whose voices are cloned, and who is responsible when infringing output is a by-product of an AI system.

As of 15 October 2026 the wave of litigation is continuing. In May 2026 the Delhi High Court released several platform defendants from the Bachchan suits after they complied, and the Jackie Shroff suit was decreed on 21 May 2026. On 6 July 2026 it modified its Sonakshi Sinha order at the request of Character Technologies, with the plaintiff's consent, so that the platform's obligation covers only her name, image and voice. On 5 October 2026 it decreed Dr Devi Prasad Shetty's suit and made its injunction against AI deepfakes permanent. The newest interim orders verified are those for Varun Dhawan (29 May 2026) and Ravi Kishan (2 July 2026), with court hearings listed for 1 October and 16 October 2026 respectively; the outcomes of those hearings have not been verified.⁹¹⁰

No Indian statute has yet been enacted on personality rights or AI likeness cloning, and no appellate court has yet laid down an authoritative test for ex parte AI personality-rights injunctions. Related government measures on deepfakes and synthetic media, including the IT Rules and MeitY advisories, are covered in other entries.