Author: Rucha Paramane
College: Dr D Y Patil Law College, Pune
LinkedIn Profile Link: ruchaparamanelinkedin
To the Point
Generative AI tools can now write essays, compose music, and paint images in seconds, raising a question copyright law was never designed to answer: can a non-human “author” hold legal rights over its output? Indian copyright law, like most jurisdictions, was built on the premise that authorship requires human intellect, skill, and judgment. With AI-generated works flooding creative industries, courts and legislatures worldwide are being forced to decide whether existing frameworks can stretch to cover machine-generated content, or whether entirely new categories of protection are needed. This article examines the current legal position in India and abroad, the doctrinal tensions involved, and the direction the law appears to be heading.
Use of Legal Jargon
A few terms are central to this discussion. “Authorship” under Section 2(d) of the Copyright Act, 1957 refers to the person who creates the work; for a “computer-generated work,” the Act attributes authorship to “the person who causes the work to be created.” “Originality” requires that a work not be copied and reflect a minimal degree of creativity or “skill and judgment,” per the Indian Supreme Court’s reasoning in Eastern Book Company v. D.B. Modak. “Sui generis protection” refers to a standalone legal regime created specifically for a new category of subject matter, distinct from existing copyright or patent law. “Training data” refers to the large corpora of text, images, or other content used to teach an AI model patterns, which raises separate questions of reproduction and fair dealing under Section 52. Finally, “moral rights” under Section 57 protect an author’s right to claim authorship and object to distortion — rights that presuppose a human author capable of having reputation and creative intent.
The Proof
The clearest indicator of the law’s current posture comes from the Copyright Act’s own text. Section 2(d)(vi) defines the author of “any literary, dramatic, musical or artistic work which is computer-generated” as “the person who causes the work to be created.” This provision predates modern generative AI by decades — it was inserted with simpler computer-assisted tools in mind — yet it remains the only statutory hook available in India for AI-generated works. Its language assumes a human “cause,” which becomes increasingly difficult to identify when an AI model autonomously generates output from a brief prompt with minimal human creative input.
In November 2020, the Copyright Office accepted an application listing an AI tool, “RAGHAV,” as a co-author alongside a human applicant for an artistic work — reportedly the first such instance globally of an AI system being named co-author in an official copyright registration. However, this registration has since been the subject of scrutiny, with the Copyright Office reportedly issuing a withdrawal notice and seeking clarification on the AI’s legal status, illustrating the institutional uncertainty surrounding the issue even where registration was initially granted.
Internationally, the position has hardened against AI authorship. The U.S. Copyright Office, in its 2023 guidance and subsequent Part 2 report on AI (2025), reaffirmed that copyright protection requires human authorship, denying registration to works generated solely by AI without sufficient human creative control, while clarifying that works combining human creativity with AI assistance may be protected to the extent of the human contribution. The UK’s Copyright, Designs and Patents Act, 1988, by contrast, already contains a provision (Section 9(3)) attributing authorship of computer-generated works to the person who made the arrangements necessary for the creation of the work — a model India’s own provision was loosely based on, though the UK has faced calls for reform given the rise of generative AI, with a 2023–2024 government consultation considering whether this approach remains fit for purpose.
On the input side, multiple lawsuits filed by authors, artists, and news organizations against AI developers (including the widely reported New York Times v. OpenAI and Microsoft litigation in the United States) allege that training large language models on copyrighted material without licence constitutes infringement, while AI companies have argued fair use protections apply. These cases remain pending and their outcomes are likely to significantly shape how training data practices are regulated globally.
Abstract
This article explores the unsettled legal terrain at the intersection of artificial intelligence and copyright law, focusing primarily on the Indian framework while drawing comparative insight from the United States and United Kingdom. It argues that India’s existing statutory provision for “computer-generated works” is inadequate to address the realities of generative AI, which can produce sophisticated creative output with minimal human direction. The article surveys key case law, regulatory responses, and pending litigation concerning both the output side (who owns AI-generated content) and the input side (whether training AI on copyrighted material infringes existing rights). It concludes that while courts continue to apply human-authorship requirements as a stopgap, meaningful reform — whether through judicial interpretation, legislative amendment, or a new sui generis regime — is increasingly necessary to balance innovation incentives with the rights of human creators.
Case Laws
Eastern Book Company v. D.B. Modak (2008) 1 SCC 1 — The Supreme Court held that copyright protection requires a “modicum of creativity” reflecting the author’s own skill and judgment, rejecting the lower “sweat of the brow” standard. This precedent is significant because it ties copyright protection closely to demonstrable human intellectual effort, a standard that AI-generated works struggle to satisfy without substantial human creative input.
Naruto v. Slater, 888 F.3d 418 (9th Cir. 2018) — Although concerning a monkey rather than AI, this U.S. case is frequently cited in AI authorship debates. The Ninth Circuit held that a macaque could not hold copyright in a “selfie” photograph because the Copyright Act’s authorship provisions apply only to human beings, reinforcing the broader principle that non-human entities lack standing to claim authorship.
Thaler v. Perlmutter, 687 F. Supp. 3d 140 (D.D.C. 2023), affirmed by the D.C. Circuit (2025) — Stephen Thaler sought to register a work created entirely by his AI system, “DABUS,” listing the AI itself as author. The U.S. Copyright Office denied registration, and the courts affirmed that the Copyright Act requires human authorship, with the D.C. Circuit upholding this position on appeal. This remains one of the most direct judicial rulings globally on whether an AI system itself can hold copyright, and the answer was an unambiguous no.
Conclusion
The law’s current answer to “can a machine own copyright” is a fairly confident no — authorship requires a human mind exercising creative judgment, and courts in India, the US, and the UK have largely held this line. But this answer increasingly sits in tension with commercial reality: AI-generated content is being published, sold, and disputed at scale, often with only a thin layer of human involvement (a prompt, a selection, a edit) standing between the user and a fully autonomous output. India’s “computer-generated work” provision, drafted for an earlier generation of technology, offers only a partial and ambiguous answer, attributing authorship to whoever “causes” the work without clarifying how much human contribution is sufficient. Until Parliament or the courts provide clearer guidance — whether by raising the threshold for human authorship, creating a separate weaker-protection category for AI-assisted works, or addressing the training-data infringement question head-on — creators, AI developers, and rights-holders will continue to operate in a zone of considerable legal uncertainty. Given the pace at which generative AI is being adopted commercially, this is an area where the law is likely to evolve rapidly over the next few years, and where India’s eventual approach will have significant implications for its creative and tech industries alike.
FAQs
Q1. Can an AI system be listed as the author of a copyrighted work in India?
No. Indian law attributes authorship of computer-generated works to “the person who causes the work to be created,” not the AI system itself. The AI tool cannot independently hold copyright.
Q2. If I use an AI tool to generate an image or piece of writing, do I own the copyright?
It depends on your level of creative input. If your contribution involves significant skill and judgment — for instance, detailed prompting, selection, arrangement, or substantial editing — you may have a stronger claim to authorship. Purely mechanical use with minimal human input is legally riskier to claim.
Q3. Is training an AI model on copyrighted books, articles, or images illegal?
This is currently being litigated and is not settled. AI companies often argue such use is permissible (akin to fair use/fair dealing), while authors and publishers argue it constitutes unauthorized reproduction. Outcomes are likely to differ across jurisdictions.
Q4. Has any country given AI systems copyright over their own creations?
No major jurisdiction currently grants copyright directly to an AI system. The closest existing models (like the UK’s computer-generated work provision) still attribute authorship to a human who arranged for the work’s creation.
Q5. What should businesses using generative AI tools do to protect their content?
Maintain clear records of human creative input (prompts, edits, selections), avoid relying solely on raw AI output for high-value content, and monitor licensing terms of the AI tools used, since some providers disclaim ownership or impose usage restrictions on outputs.
