Author: Mahak Chatkele
Rabindranath Tagore university
Introduction
He typed fourteen words. The machine did the rest.
A nineteen-year-old from Pune entered a digital art competition in 2024 and won. The judges called it striking, layered, emotionally precise. What they did not know what he did not tell them until weeks later was that he had not drawn a single line. He had typed a prompt into an AI image generator and watched it produce, in seconds, what would have taken a trained artist days.
The internet had opinions. The competition had questions. But the law the one that was supposed to decide who owns what was created had nothing. No provision. No precedent. No answer.
This is not an isolated incident. It is a pattern. Across India, designers are generating logos, musicians are cloning voices, writers are producing content all with artificial intelligence, all at a scale and speed that no human hand could match. And behind every piece of AI-generated work sits a question that the Copyright Act 1957 was never written to answer: who is the author when the author is a machine?
India’s copyright law was built for a world where creation required human hands, human choices, and human intent. It assumed that behind every painting was a painter, behind every song a songwriter, behind every book a writer. It did not imagine a world where fourteen words typed into a phone could produce a prize-winning artwork in under a minute.
That world is here. The law is not.
Abstract
Artificial intelligence has collapsed the distance between imagination and creation. A prompt typed in seconds can produce a painting, a song, a screenplay, or a logo that would have taken a trained professional weeks to complete. But Indian copyright law was built for a world where authorship required human skill, human labour, and human intent and it has not caught up.
The Copyright Act 1957 defines an author as the person who creates a work. It does not define what happens when the creator is an algorithm. It does not address whether a prompt constitutes creative input sufficient for ownership. It does not say who owns output when the AI that generated it was trained on millions of copyrighted works without their creators’ consent.
This article examines AI-generated works through the lens of the Copyright Act 1957, the Information Technology Act 2000, and the Digital Personal Data Protection Act 2023. It analyses the legal vacuum India currently inhabits, compares approaches taken by the United States, United Kingdom, and China, and argues that India urgently needs a legislative framework that answers the question its courts have not yet been asked to decide who owns what a machine creates?
To the Point
The law loves to celebrate human creativity. It has never learned to regulate the machine that replaced it.
Artificial intelligence is not arriving in India it is already here, already working, already generating content that is being sold, submitted, published, and contested. Graphic designers in Mumbai are losing clients to AI tools that produce brand identities in minutes. Musicians in Chennai are finding their voices cloned on platforms without their knowledge or consent. Writers across the country are watching their styles replicated by algorithms trained on their own published work.
And behind every one of these situations is a Copyright Act that asks the same question it has asked since 1957 who is the human author? In a world where the author is increasingly a machine, that question produces only silence.
The problem is not that AI creates. The problem is that Indian law has not decided what to do when it does. There is no definition of AI-generated work in any Indian statute. There is no category for it in the Copyright Office’s registration system. There is no case law that directly addresses it. There is no parliamentary committee currently examining it with any urgency.
What exists instead is a patchwork of provisions designed for a different era. Section 13 of the Copyright Act grants protection to original literary, artistic, musical, and dramatic works. Section 2(d) defines author for each category painter, photographer, composer but contains no entry for AI. Section 52 provides fair dealing exceptions but was never designed to address training data scraped from millions of copyrighted sources. Section 51 defines infringement but cannot easily capture a situation where no single work is copied but thousands are absorbed.
The celebration of creativity happens in public. The replacement of creators happens in private. And the law the one meant to stand between human imagination and machine efficiency is still reading from a document written when computers filled entire rooms.
The works exist. The ownership does not.
Legal Jargon
AI-Generated Work — Content including text, image, audio, or video produced by an artificial intelligence system with little or no direct human creative input beyond a prompt or instruction.
AI-Assisted Work — Content where a human uses AI as a creative tool — similar to using Photoshop or a camera but where the core creative decisions remain human. Indian law is more likely to protect this category than purely AI-generated work.
Prompt Engineering — The act of crafting input instructions to guide an AI system toward a specific creative output. The central legal question is whether this constitutes sufficient “skill and labour” to establish authorship under Indian copyright law.
Training Data — The vast collection of text, images, audio, and video much of it copyrighted used to train AI systems. Whether scraping and using this material constitutes infringement under Section 51 of the Copyright Act is one of the most contested questions in global IP law today.
Originality Standard — Established in Eastern Book Company v. D.B. Modak, Indian copyright requires a “modicum of creativity” beyond mere mechanical reproduction. Whether AI output meets this standard and whether a human prompt constitutes creative input remains legally unresolved.
Computer-Generated Work — A category recognised explicitly under UK copyright law, where authorship is assigned to the person who makes the arrangements necessary for the work’s creation. India has no equivalent provision.
Moral Rights Under Section 57 of the Copyright Act, authors hold the right to claim authorship and object to distortion of their work. Whether these rights attach to AI-generated output and who holds them is entirely unaddressed in Indian law.
The Proof
The data does not lie even when the image was never real.
According to a 2024 Adobe study, 38% of Indian creators under thirty have used generative AI for commercial work. Of those, over 70% did not know who legally owned the copyright to what they had generated. They created. They sold. They moved on. The legal question sat unanswered beneath every transaction.
The Copyright Office of India has not issued a single guideline on AI authorship as of 2026. There is no registration category for AI-generated work. There is no official position on whether a prompt constitutes creative input. A creative economy worth billions of rupees is operating entirely without legal clarity on its most fundamental question who owns what is being made.
The scale of the problem compounds daily. One person with access to a generative AI tool can produce hundreds of images, articles, or musical compositions in a single afternoon. One company can train a model on the entire catalogue of Indian classical music and release derivative works the next morning. The original artists whose work trained that model will never know, never consent, and never be compensated.
What makes AI-generated content uniquely legally complex is its origin. Unlike a photograph that copies a painting, AI output does not directly reproduce any single source. It absorbs patterns from thousands of sources simultaneously and generates something new new enough to avoid traditional infringement claims, but built entirely on the creative labour of human beings who were never asked and never paid.
The proof is not just in the statistics. It is in every designer who found their distinctive style replicated by an algorithm. Every writer who discovered their prose rhythm in a chatbot’s output. Every musician who heard their voice cloned on a platform without permission.
And had no section of law to point to.
Case Laws
Indian courts have not yet delivered a landmark judgment specifically on AI authorship in part because the Copyright Office has not yet registered such a work, and in part because the legal profession is still catching up to the technology. But the principles that will govern such a judgment already exist in Indian jurisprudence, and the world’s major copyright jurisdictions have already begun choosing sides.
Eastern Book Company v. D.B. Modak (2008)
The Supreme Court held that copyright requires a “modicum of creativity” not mere mechanical or trivial effort. The judgment rejected the “sweat of the brow” doctrine, establishing that labour alone without creative input is insufficient for copyright protection. Applied to AI, the question becomes whether a human prompt the instruction that produces the work constitutes creative input, or whether the creativity resides entirely in the machine’s output, where no human authored anything.
R.G. Anand v. Delux Films (1978)
The Supreme Court distinguished between ideas and their expression copyright protects expression, not ideas. AI models are trained on millions of copyrighted expressions. Whether absorbing expression to generate new expression constitutes infringement under Section 51 — or falls within fair dealing under Section 52 — is the training data question Indian courts have not yet answered.
Zarya of the Dawn — US Copyright Office (2023)
The US Copyright Office partially cancelled copyright registration for a graphic novel whose images were purely AI-generated, holding that only the human-authored text and the selection and arrangement of images qualified for protection. The purely machine-generated images did not. The US position is clear human authorship is required, and a prompt alone does not provide it.
Section 9(3) Copyright Designs and Patents Act 1988 — United Kingdom
The UK explicitly addresses computer-generated works, granting authorship to the person who makes the arrangements necessary for the work’s creation. This is currently the most workable model for jurisdictions grappling with AI authorship it identifies a human responsible party without requiring traditional creative input.
Beijing Internet Court — China (2023)
A Chinese court granted copyright protection to an AI-generated image, holding that the user’s prompt demonstrated sufficient intellectual input and creative choice to constitute authorship. China’s approach is the most permissive globally and the most likely to encourage AI innovation at the cost of traditional creator protection.
Together these cases demonstrate that every major jurisdiction is making a choice India has not yet made. The gap between what is being created and what the law protects grows wider every day courts and legislatures remain silent.
Analysis
The legal vacuum around AI-generated works in India is not accidental. It is the product of a legislative framework that was last meaningfully updated in 2012 before generative AI existed in any publicly accessible form and a Copyright Office that has chosen caution over clarity in the absence of judicial direction.
The Authorship Problem
Section 2(d) of the Copyright Act defines author differently for different categories of work the painter for artistic works, the composer for musical works, the photographer for photographs. There is no entry for AI-generated work because the category did not exist when the provision was drafted.
Two interpretations are possible under the current text. The first that the person who types the prompt is the author, because they “cause” the work to be created through their instructions. The second that no copyright attaches to purely AI-generated work at all, because the Act’s definition of author always contemplates a human exercising creative judgment, and a prompt is not creative judgment in the way the Act understands it.
Neither interpretation is clearly correct under existing law. Both leave the creator whether human prompter or AI system in legal uncertainty. That uncertainty has real economic consequences in a country where the creative economy is increasingly digital and increasingly AI-assisted.
The Training Data Problem
Equally unresolved is the question of what happens to the human creators whose work trained the AI in the first place. Every major generative AI model was built on vast quantities of copyrighted material images, text, music, code scraped from the internet without the consent of the people who created it.
Section 52 of the Copyright Act provides fair dealing exceptions for research, private study, criticism, and review. Whether training an AI model constitutes “research” within the meaning of Section 52 has never been tested in an Indian court. If it does not and there are strong arguments that commercial AI training is not what Parliament had in mind when it drafted the research exception then every major AI company operating in India may be exposed to infringement liability on a scale the legal system has no mechanism to process.
The Way Forward
India needs three things none of which require waiting for the Supreme Court to accidentally encounter the right case.
First, an amendment to Section 2(d) adding a definition for AI-generated and AI-assisted works, establishing that a human who exercises substantial creative control over an AI-generated work may claim authorship in it while purely prompt-generated work without meaningful human creative input receives no copyright protection.
Second, a sui generis protection regime for AI-generated works that receive no copyright a shorter-term producer’s right, similar to the protection given to sound recordings, that rewards investment in AI creation without pretending a machine is an author.
Third, mandatory disclosure and transparency requirements for AI training data building on the DPDP Act 2023’s framework requiring AI companies to maintain records of training sources and providing Indian creators with opt-out rights and compensation mechanisms.
These are not radical proposals. They are versions of what the UK, EU, and China have already implemented or are actively legislating. India does not need to invent a new model. It needs to choose one and choose quickly, before the creative economy builds itself entirely on a legal foundation that does not exist.
Conclusion
He typed fourteen words. The machine made something extraordinary. And the law had nothing to say about who owned it.
AI-generated works are not a future problem waiting to arrive. They are a present reality being created, sold, and contested right now in competitions, in agencies, in courtrooms that have not yet received the right case, and in a Copyright Office that has not yet issued the right guideline.
India’s creative economy is too large, too diverse, and too significant to be left operating in legal silence. Every designer who loses a client to an AI tool deserves a law that decides whether that tool’s output can be protected. Every musician whose voice is cloned deserves a provision that names what happened to them. Every writer whose work trained a model without their consent deserves a framework that makes that consent necessary.
The Copyright Act 1957 was written for human hands. The hands have not disappeared but they are no longer the only ones creating. And a law that only sees the human hand will miss everything the machine is building beside it.
Your prompt is yours. Your idea is yours. Your creative choices are yours.
But until Indian law decides what that means when a machine does the making the art will keep being generated, the ownership will keep being contested, and nobody will be answerable to it.
FAQ
Q1. Can I register copyright for an image I created using an AI tool like Midjourney or Adobe Firefly?
Under current Indian law there is no clear answer the Copyright Office has issued no guidelines on AI-generated works and has no dedicated registration category for them. Globally, the trend is toward protecting AI-assisted works where substantial human creative input is demonstrated, while denying protection to purely prompt-generated output. Until India legislates or the Copyright Office issues guidelines, registration attempts for purely AI-generated work are likely to remain in limbo.
Q2. If an AI model was trained on my artwork without my permission, do I have a legal remedy in India?
Potentially, but it is untested. If the training involved reproduction of your specific copyrighted work without falling within Section 52’s fair dealing exceptions, you may have an infringement claim under Section 51. The challenge is establishing that your work was specifically used, proving the connection between your work and the model’s output, and navigating a legal system that has no precedent for this type of claim. This is an area where Indian IP law urgently needs both judicial clarity and legislative intervention.
Q3. Which country’s approach to AI copyright is India most likely to follow?
The UK model — which assigns authorship of computer-generated works to the person who makes the arrangements necessary for their creation is the most compatible with Indian copyright jurisprudence and the most practically workable. However, India may also draw from the EU’s approach to training data transparency and China’s willingness to recognise substantial prompt input as creative authorship. The most likely outcome is a hybrid framework if and when Parliament chooses to act.
