Bombay High Court orders removal of AI-morphed images of Shilpa Shetty
The Bombay High Court sees fit to give an order for the ‘deletion/removal of morphed/deep fake videos produced through artifices of Artificial Intelligence using the image/likeness of, and purportedly voice/mannerisms of, Shilpa Shetty.’ It appears to indicate the gravity with which the Indian courts are treating the use of Artificial Intelligence for sexualized/misleading impersonations. “It’s extremely disturbing and shocking/abhorrent.”
The following is a lengthy, 3000-word style explanation of what the order is, its significance, and what it reveals about where the laws of India may be going on deepfakes and “personality rights”:
“Personality rights” encompasses rights of privacy, rights of publicity, and certain
1) What is the Shilpa Shetty case about?
The Basic Complaint
One such case is that of actress Shilpa Shetty, who filed a case in the Bombay High Court, claiming that morphed/deepfake pictures and videos were being generated using Artificial Intelligence, featuring her, which were inappropriate, unacceptable, andए humiliating for her. Cases also include morphing/imitation of her voice, her mannerisms, which were further being used in books/merchandise, among other uses, implying that it was a problem not just related to reputation but also potential profitability.
What the court did (interim relief)
A vacation bench of the Bombay High Court (reported as the case Justice Adviat M. Sethna) had granted “interim relief—a brief, pressing, and temporary measure to prevent harm to the plaintiff during the pendency of the suit.” The important aspect of the direction was to “delete/remove/take down the links/posts/webpages carrying the morphed/AI-generated images and to refrain from spreading the same in
Why ‘Interim’ Matters
First of all, the interim order is not the final order on all matters of law that arise between the spouses during the divorce process.
It nevertheless carries considerable authority since it:
acknowledges prima facie (at first look) wrongdoing,
recognizes urgent, ongoing harm, and
forces platforms/intermediaries to act promptly.
In the case of “deepfakes,” the quicker the better, since once the video gets out, copies proliferate, and the videos just keep popping back up. That’s why the courts rely on rapid takedown notices and, in some cases, “dynamic injunctions” (removing the URLs of the day, as well as the current ones).
2) The tone of the courts: “disturbing and shocking” is not colloquial language
The usual tone of the courts is measured. When the bench employs “extremely disturbing and shocking” and “abhorrent,” it is actually accomplishing two things:
Recognizing the severity of the damage
Deepfakes are no mere “edited photos.” Deepfakes can construct an entire fiction of reality, using one’s face/body/voice for harassment, sexual degradation, blackmail, or reputation demolition.
Indicating urgency to the intermediaries
Platforms can move most quickly in response to a court order if that order is clear, definitive, and forceful. The use by the court of stronger language reinforces the expectation of prompt compliance by social media and web hosts.
3) What is the concept of “personality rights,” and why are celebrities leveraging this concept?
Explanation in simpler terms
“Personality Rights,” at times synonomous with “Publicity Rights,” are those personal interests entitled to protection regarding:
their name
image/
voice
signature
likeness
characteristic features of their personality (sometimes even style and catchphrases)
The concept is: You can’t commercially gain from another person’s identity without his/her consent, especially if there is any form of damage or deception involved.
Why this is the go-to legal route
This legal route is
Currently, India does not have a “deepfake law” by itself. As such, the plaintiffs will rely upon a combination of the following:
Right to privacy and dignity (right protected under constitution/Article 21-based argument through case law): This right
Tort of Passing Off (Misrepresentation Constituting Deception)
Copyright issues (infringement of images/m
Similar rights as Trademarks: Persona Protection
Also read: Trademark
Obligations of Intermediaries and Takedown
In recent times, courts are more open to upholding celebrities’ rights under the auspices of “PERSONALITY/Publicity rights,” especially in IP matters, as the injury occurs immediately and is readily visible.
4) Who were the defendants, and why is “John Doe” important?
“Known + unknown” defendants
According to reports, these claims included:
I.
social media platforms/intermediate
AI-connected entities or websites
and unknown persons (“John Doe”/“Ashok Kumar” defendants
This is very important, considering that original uploaders can be impossible to trace rapidly within deepfakes, while re-uploaders are found everywhere.
What a John Doe order can accomplish
A John Doe order
John Doe injunctions give the court the benefit of extending protection to unknown offenders, meaning the court order can be issued to the platform/host prior to identifying the alleged offender. It can be summarized as follows in the words of Big Chill:
“Stop whoever is doing this, and stop wherever it appears.”
“This mechanism is widely used in piracy cases and is now increasingly used in deepfakes/personality-rights infringement because the pattern of distribution is the same—very fast, anonymous transmission, and a focus on copies.”
5) Why courts describe deepfakes as a “dignity” issue and not merely “defamation”
“Defamation” is only one slice of the
“The law of defamation is concerned with reputation. However, deepfakes tend to go beyond that
The sexualized morph may be a form of gender humiliation:
“It may also amount to non-consensual sexualized impersonation:
It can cause safety issues (stalking, harassment)
It may adversely affect mental health
It may affect career development chances
Thus, the importance of privacy and dignity terminology arises. In this particular situation, the reporting emphasizes how the court has outlined the concept of personality rights as a protective tool against privacy, especially women.
The ‘commodification’ argument
In this case, in the words of the lawyer of the actor involved in the production, “As quoted in the Boston Globe, ‘One of the things to keep in mind is that the identity of a woman should not be considered a commodity.’ Indeed, this is a crucial underlying issue in the use of deepfakes—that it makes the individual involved a
6) What is the meaning of ‘takedown’
It is typically necessary to have the following items to
Deletion of particular posts/URLs
Precluding acccess to specific web pages
De-indexing (occasionally) from search results
Preventing Re-Uploads ( harder, yet called for
“Real-world challenge: re-uploads
The very same material can then reappear after the take-down:
mirrored sites
alternate spellings/
slightly edited versions to avoid being detected
So, in contemporary internet injunction cases, courts do sometimes allow plaintiffs to submit new URLs to sites for removal without requiring the courts to be involved each time (a “dynamic” notice and takedown procedure). Whether such specific procedure was available in this particular case is a question of specific textual analysis of what the notice states, but the overall guidance to delete links/posts/websites is indeed reported.
7) What is likely doing the heavy lifting for the applicable law?
Given the style in which Indian “personality rights” litigation tends to be framed and in the manner in which this case has been presented in the reports:
- “The Applicant’s claim in this case
A) Privacy and Dignity
There is an increasing recognition that courts see intimate, manipulated, or sexualized imitation as an attack on privacy and human dignity itself rather than simple reputation.
B)Misrepresentation/passing
If the content implies endorsement, association, or “this is really her,” it may be considered deceptive.
C) IP/IP rights + unlawful use
Photos/videos and manipulation works can give rise to copyright problems (particularly in the use/reproduction of original works).
D) Intermediary responsibility and take-down rules
Platforms may need to meet those court requirements quickly and stop the spread of information.
In reporting, the dimension is clearly defined as a personality right, focusing on privacy/dignity.
8) Reasons why this particular order is significant and not specific to this particular individual
“Deepfakes are moving from the realm of ‘novelty’ to
Artificial intelligence achieves high quality in manipulation at a low and rapidly decreasing cost:
face swaps
voice cloning
synthetic nude production
“scene insertion” within pornographic material
fabricated interviews and political statements
A tough takedown notice issued by courts leads to the creation of:
deterrent (at least against visible and monetized activity)
compliance pressure on platforms
precedent value for future victims (non-celebrities included)
It assist to conceptualise ‘personality rights’ in the age of AI.
“Personality rights” originally concerned advertising/endorsements and merchandise. “Deepfakes” extend “personality rights” to:
identity integrity (don’t make me “do” things I didn’t do)
sexual privacy (do not use my likeness for sexual purposes)
voice and mannerisms (identity markers aside from facial features)
This is quite a dramatic change—and courts are now crafting orders that incorporate these changes.
9) Implications for platforms and content creators in India
For platforms
Courts are increasingly requiring:
Quick response to valid court directions
robust reporting and takedown mechanisms
improved detection for re-uploads of the same harmful media
Platforms, if they move slowly, can receive stricter judicial instructions in later litigations.
For Creators & Marketers
The “AI can generate it” excuse is not applicable in cases where:
you used someone’s persona without their permission
you created sexualised impersonations
you implied endorsement/association
Nevertheless, even parody claims can go down in “laughingstock” with regard to obscene, harassing, or plausibly misleading communication.
10) What the victims, whether celebrities or otherwise, can learn from this incident
If one’s morphed/deepfake content is out, then the process is as follows:
Preserve Evidence
screenshots, URLs, timestamps.
Screen recordings (including URL and date and time)
Report to platforms
utilize impersonation or non-consensual intimate imagery policies when relevant
Send legal notices
to intermediaries, to the host, to domain registrars if necessary
Report to law enforcement
especially where it’s obscene/sexualized, extortionate, or threatening
Apply to court for urgent relief
Especially in cases where the information is viral or disseminated by multiple sources
It is obvious that the courts can certainly act swiftly in cases in which there is obvious damage being caused.
2 11) Bigger picture: India’s deepfake moment This comes in a time when public discourse in India has transitioned from “AI is cool” to “AI can ruin lives.” Deepfakes are not as common anymore, and the justice system is addressing this through: privacy and personality rights jurisprudence, IP/commercial cases takedown and John Doe injunctions, and the growing pressure on the platforms to consider synthetic impersonation a form of high-harm content. The Shilpa Shetty case is important, first, because it brings the problem of deepfake abuse well and truly within the ambit of rights, dignity, and identity, and not simply the realm of PR challenges facing the stars of the entertainment industry. The Indian Express 2 “Times of India” 2 12) What will happen in this case? Generally, following relief regarding: defendants/platforms file responses (or compliance reports), plaintiff may submit additional URLs on which it prefers its content removed by the “the court may enlarge directions (if fresh matter arises),” The issue is thus referred towards a further hearing and perhaps a final injunction or settlement. If the removal is done quickly, there may be fewer appearances of the content in mainstream search results—but it may very well be seen on smaller sites or when it is re-posted or encrypted-shared—and so “dynamic” removal becomes necessary in such circumstances.
