For a voice artist, the asset is the instrument. Your voice is the product, the brand and the livelihood, and for the first time it can be copied. AI voice cloning can take a few minutes of your recordings and generate new lines you never spoke, in your timbre, your style, your signature delivery. In India, where a single dubbing performance can be the seed for versions in twenty languages, that is not a distant worry. It is a present threat to how the whole craft gets paid. This guide explains what Indian law actually protects today, what the recent court rulings do and do not do for you, and the concrete steps, mostly contractual, that put a wall around your voice. It is general information on how the law stands, not legal advice; for your own contract, talk to a lawyer.
The threat is already in the studio
This is not speculative. Voice artists in India report the work drying up. The Association of Voice Artists of India has described a performer who once did 15 to 20 projects a month now doing six or seven, as text-to-speech eliminates narration work entirely. India’s roughly 20,000 freelance voice artists sit in an informal sector with no SAG-AFTRA-style binding protection. The economics that made Indian dubbing a gold mine, one story retold across many regional languages, is exactly what AI now exploits: clone the performance once and deploy it across every language without re-hiring or re-paying the artist. Producers have begun arguing, in the words one artist reported, that they are not airing your voice, merely using your performance and duplicating it. Major multi-language films have already used AI voice cloning. The displacement is underway, and the threat is not only economic. As one artist put it, if a clip of hate speech surfaces in your voice tomorrow, you will be the one horrified.
What the courts have done, and its limits
Indian courts have moved faster than the legislature, and consistently in the artist’s favour, in a series of high-profile cases:
- Anil Kapoor v. Simply Life India (Delhi High Court, 2023) was the first Indian order to name AI and deepfakes, granting a broad injunction protecting the actor’s name, voice, likeness and dialogue delivery against AI and face-morphing tools used for commercial gain. It established voice as a protectable facet of personality rights.
- Arijit Singh v. Codible Ventures (Bombay High Court, 2024) was the first Indian decision squarely on AI voice cloning. The court protected the singer’s voice, vocal style, mannerisms and persona, and held that making AI tools available to convert anyone’s voice into a celebrity’s without consent violates personality rights. The judge noted how exposed performers are to unauthorised generative content.
- Jackie Shroff v. The Peppy Store (Delhi High Court, 2024) restrained an AI chatbot and other tools from exploiting the actor’s persona, voice and likeness without consent.
- Asha Bhosle v. Mayk Inc. (Bombay High Court, 2025) and Kumar Sanu v. Jammable (Delhi High Court, 2025) extended the line directly to AI voice-conversion platforms, ordering takedowns and disclosure of the people behind the cloned audio, and reaffirming that a celebrity’s voice cannot be commercially replicated by technological means without authorisation.
Read these correctly. Every one is an interim or ad-interim injunction, not a final judgment on the merits, so the law is being established case by case, not settled. And every winning party is a famous artist. Personality rights, the tool doing the work in these cases, require you to first prove commercial value in your own identity. A jobbing dubbing artist or a session voice may not clear that bar. So the headlines are encouraging, but do not assume that because Arijit Singh won, you automatically have the same remedy. The celebrity gate is the single most important caveat in this whole subject.
What the law actually gives an ordinary voice artist
India has no specific AI statute yet, so protection is a patchwork, and the pieces are uneven:
- Personality and publicity rights. Judge-made, the strongest tool in the cases above, but gated on proving you are a recognisable, commercially valuable identity. Powerful if you qualify, of little help if you do not.
- Performer’s rights under the Copyright Act (Sections 38, 38A, 38B). These give performers, including singers and actors, exclusive rights over the fixation, reproduction and broadcast of their actual recorded performance, plus moral rights, for 50 years. The catch is crucial: they protect the specific recording, not the timbre of your voice. An AI that generates a wholly new performance from a clone of your voice may not be reproducing any protected recording, so this may not catch it. That gap is untested in a final Indian judgment.
- Moral rights (Section 57) and passing off. Moral rights attach to defined works, not to a vocal persona, so sound-alikes tend to fall outside. Passing off, a common-law tort, can restrain commercial use that misrepresents your endorsement, but needs misrepresentation, goodwill and damage.
- IT Act and intermediary takedown. The basis for the takedown and disclosure orders against marketplaces and platforms in the recent cases, useful for getting cloned audio pulled down once it exists.
The honest summary: for a non-famous voice artist, litigation is uncertain and expensive, and the one lever that works regardless of fame is the contract. So that is where to spend your energy.
The contract is your real shield: clauses to insist on
Treat every engagement as if your voice could be cloned from it, because it can. Build an AI rider into your agreements and refuse work that will not accept reasonable terms. The clauses that matter:
- No-AI-training. Your recordings may not be used to train, fine-tune or develop any AI or machine-learning voice model.
- No synthetic recreation or cloning. Bar the creation of any synthetic voice, digital double or voice clone derived from your performance.
- Separate consent and separate pay. Any AI or synthetic use requires fresh written consent and a distinct fee, never bundled into the session rate. This is the clause that keeps cloning from being a free bonus to the buyer.
- Scope and medium limits. Name the exact medium and use, and bar repurposing into others without renegotiation. Critically for India, require per-language payment, so a single recording cannot be deployed across regional languages for free.
- Term limits. Fixed start and end dates. Strike any “in perpetuity” or “throughout the universe” language and renegotiate at term end.
- No third-party transfer. The client cannot sell, license or hand your recordings to outside AI developers.
- Revocation and credit. A right to withdraw AI consent and require cessation if misused, plus an attribution requirement.
Learn the red-flag words and refuse them when they appear without protection and separate pay: simulation, synthesization, digital double, machine learning, machine training. Beyond the contract: ask for full project specifications before you audition, so you know whether AI is involved; document any refusal in writing; register significant performances to strengthen any later performer’s-rights claim; and keep evidence of your public recognition, which is what builds the celebrity-status case if you ever need to litigate. If your voice is misused, the route is an intermediary takedown notice and an injunction action, as in the recent rulings.
The multi-language trap, and collective action
For Indian voice artists, one clause matters more than any other, and it is the one producers most want to skip: per-language payment. India’s linguistic range was the dubbing industry’s gold mine, because one story sold across twenty languages meant twenty engagements. AI inverts that. Record once, clone the voice, and deploy it across every regional language without re-hiring or re-paying anyone. A voice that earns a theatrical dubbing fee for one language can be stretched across all of them for free unless your contract forbids it. If you take nothing else from this guide, take this: never sign away “use across all languages” without separate payment for each, because that single line is where most of the craft’s future income leaks out.
No individual fixes this alone, which is why collective action matters. The Association of Voice Artists of India has been issuing circulars, running awareness forums, and pressing for the three things that protect the craft: consent, credit and fair pay. India has nothing like the binding studio agreements that US performers won through their union, so for now the protection you can rely on is the one you negotiate yourself, backed by whatever collective pressure your peers can muster. Worth noting too: AI is not only a threat here. Cloning studios and AI-dubbing work are creating new roles for artists who learn to direct, supervise and license synthetic voices on their own terms. The artists who treat AI as a tool they control, rather than a wave that hits them, are the ones writing the better contracts.
Why a verified record matters here too
The thread running through all of this is proof. The artists who win are the ones who can show their work, their recognition and their terms. A clear, verified record of what you have voiced, for whom, under what consent, is both leverage in a negotiation and evidence if a clone surfaces. This is the same logic that protects pay across the industry, and it is why we keep returning to verified credits as infrastructure rather than vanity. AI does not erase the value of a real human voice and a real human judgment. It does make it essential to document and defend them.
TUAP is free for crew and built so your work and your terms travel with you. Claim your page, record your credits and the consent that came with them, and present your craft on proof rather than promises. As AI reshapes which roles survive, the artists who control their record and their contracts are the ones who keep the work. For the wider picture of which creative roles AI is changing, read will AI replace video editors in India.