ABSTRACT
Generative artificial intelligence systems can now produce literary, artistic, and musical output with minimal human direction, straining a copyright framework built on the assumption that a work has an identifiable human author. India is unusual among major jurisdictions in already possessing a statutory hook for non-human creation: Section 2(d)(vi) of the Copyright Act, 1957 vests authorship of a "computer-generated" work in "the person who causes the work to be created." This article asks whether that thirty-year-old provision, inserted before generative AI existed, can coherently accommodate AI-authored works, and how India's approach compares with the human-authorship models adopted in the United States and the United Kingdom. Using doctrinal research grounded in the Copyright Act, the Copyright Office's contradictory handling of the "RAGHAV" AI application, the Supreme Court's originality standard in Eastern Book Company v. D.B. Modak, the U.S. Court of Appeals' 2025 decision in Thaler v. Perlmutter, and the UK's Section 9(3) of the Copyright, Designs and Patents Act, 1988, the article finds that Section 2(d)(vi) is textually capable of covering generative-AI output but has never been authoritatively construed by an Indian court, leaving registration practice inconsistent and commercially uncertain. The comparative analysis shows India occupies a middle position between the United States' strict human-authorship rule and the United Kingdom's deeming provision, without the interpretive clarity either jurisdiction has developed. The article concludes that Parliament should amend the Copyright Act to define "computer-generated work," clarify the applicable originality threshold, and empower the Copyright Office to issue binding registration guidelines.
INTRODUCTION
Background
Generative AI tools such as text-to-image and large language models can now produce artistic and literary output from a short prompt, with the system rather than the human contributing most of the expressive detail. Copyright law across jurisdictions has traditionally assumed that every protectable work traces to an identifiable human author, a premise now tested by software capable of autonomous or near-autonomous creation.[1] India is distinctive because its Copyright Act, 1957 already contains a provision addressing "computer-generated" works inserted by the Copyright (Amendment) Act, 1994, decades before generative AI became commercially available which vests authorship in "the person who causes the work to be created."[2] Whether this provision can bear the interpretive weight of contemporary generative AI became a live controversy when the Indian Copyright Office, in November 2020, registered the AI application "RAGHAV" as co-author of the artwork "Suryast," only to issue a withdrawal notice roughly a year later questioning the AI tool's legal status.[3]
Research Problem
The Copyright Office's contradictory handling of the RAGHAV registration reject, grant, then attempt withdrawal reveals that Section 2(d)(vi) has never been authoritatively construed by an Indian court in the context of generative AI, leaving unresolved whether an AI system can be named as author or co-author, what "causes the work to be created" means when a human merely supplies a prompt, and what standard of originality a substantially AI-generated work must meet.[4] This uncertainty affects every business, publisher, and creator now relying on generative tools, since it is unclear whether the resulting output is protectable at all, and if so, who owns it.
Research Questions
Does Section 2(d)(vi) of the Copyright Act, 1957 extend to works generated by contemporary generative AI systems, and if so, on what interpretation of "the person who causes the work to be created"?
What standard of originality applies to a substantially AI-generated work under Indian copyright doctrine?
How does India's statutory approach compare with the human-authorship requirement in the United States and the deeming provision in the United Kingdom?
Is legislative or judicial clarification the more appropriate route to resolving the uncertainty exposed by the RAGHAV registration?
Objectives
This article aims to: (a) analyse the text and legislative history of Section 2(d)(vi) against generative-AI fact patterns; (b) examine the Copyright Office's RAGHAV decisions and the Supreme Court's originality jurisprudence for their bearing on AI-generated works; (c) compare the Indian position with the United States and United Kingdom frameworks; and (d) recommend specific statutory and administrative reforms.
Methodology
The article follows doctrinal legal research, examining the Copyright Act, 1957 and its 1994 amendment, Copyright Office registration records concerning RAGHAV, the Supreme Court's judgment in Eastern Book Company v. D.B. Modak, the U.S. Court of Appeals for the D.C. Circuit's decision in Thaler v. Perlmutter, the UK Copyright, Designs and Patents Act, 1988 and Nova Productions Ltd v. Mazooma Games Ltd, supplemented by professional and academic commentary.[5]
Scope
The article is confined to authorship and originality questions concerning copyright in AI-generated literary, dramatic, musical, and artistic works. It does not address patent inventorship, AI training-data infringement claims, or criminal liability, except where incidentally relevant.
MAIN BODY
I. THE STATUTORY TEXT: SECTION 2(D)(VI) AND ITS LIMITS
Section 2(d) of the Copyright Act, 1957 defines "author" differently for each category of work the artist for an artistic work, the person taking the photograph for a photograph, the producer for a cinematograph film or sound recording and, under sub-clause (vi), inserted by the Copyright (Amendment) Act, 1994, provides that for "any literary, dramatic, musical or artistic work which is computer-generated," the author is "the person who causes the work to be created."[6] Unlike the United States and, historically, the United Kingdom pre-1988, Indian law therefore already contemplates authorship without direct human execution of the work. The difficulty is that the term "computer-generated" is nowhere defined in the Act, and the provision was drafted in 1994 with software tools that mechanically executed human instructions in mind not systems capable of generating unpredictable, non-deterministic output from a short natural-language prompt.[7] Applied literally, "the person who causes the work to be created" could denote the AI developer, the deployer of the AI service, or the end user who supplies the prompt, and the Act offers no guidance for allocating authorship among these candidates when their contributions differ so substantially from the traditional idea of authorial execution.
II. THE RAGHAV CONTROVERSY: ADMINISTRATIVE PRACTICE WITHOUT DOCTRINE
The practical consequences of this gap surfaced in the Copyright Office's handling of applications filed by Ankit Sahni for the artwork "Suryast," generated using the "RAGHAV" (Robust Artificially Intelligent Graphics and Art Visualizer) application.[8] An initial application naming RAGHAV as sole author was rejected. A subsequent application naming Sahni and RAGHAV as co-authors was registered in November 2020 apparently the first instance globally of an AI tool being registered as a copyright co-author but the Copyright Office issued a withdrawal notice roughly a year later, asking Sahni to clarify RAGHAV's legal status, without ever articulating the statutory basis for either the original grant or the subsequent withdrawal.[9] No court has yet resolved the resulting dispute. The episode demonstrates that Section 2(d)(vi) is being applied, if at all, inconsistently and without reasoned interpretation: the Copyright Office has neither confirmed that an AI system can be a co-author, nor explained why, if it cannot, the initial registration was permitted at all. For applicants and rights-holders, this leaves registration outcomes effectively unpredictable.
III. THE ORIGINALITY THRESHOLD: EASTERN BOOK COMPANY V. D.B. MODAK AND ITS APPLICATION TO AI OUTPUT
Even where authorship can be assigned to a human "person who causes the work to be created," the work must still satisfy the Act's originality requirement under Sections 13 and 14. The Supreme Court's judgment in Eastern Book Company v. D.B. Modak calibrated this standard for India, rejecting both the low "sweat of the brow" test and the higher American requirement of novel, non-obvious creativity, and adopting an intermediate "skill and judgment" standard under which a work must reflect more than trivial or mechanical effort but need not be wholly novel.[10] Applied to generative AI, this standard raises an unresolved question: does a short natural-language prompt reflect the "skill and judgment" the Supreme Court required, or is prompting closer to the trivial, mechanical contribution the Court held insufficient in relation to routine copy-editing?[11] Because no Indian court has yet applied the Eastern Book Company standard to an AI-generated work, prompt-based creation occupies the same doctrinal vacuum as the authorship question itself a work might satisfy Section 2(d)(vi)'s authorship test while failing the originality threshold, or vice versa, and current law provides no way to predict which.
IV. COMPARATIVE ANALYSIS: HUMAN AUTHORSHIP IN THE UNITED STATES AND THE DEEMING PROVISION IN THE UNITED KINGDOM
The United States has taken the opposite approach to India, insisting on human authorship as an absolute precondition for copyright protection. In Thaler v. Perlmutter, the U.S. Court of Appeals for the District of Columbia Circuit affirmed the Copyright Office's refusal to register an artwork listing an AI system, the "Creativity Machine," as sole author, holding that the Copyright Act "requires all eligible work to be authored in the first instance by a human being"; the Supreme Court denied certiorari in March 2026, leaving that human-authorship rule undisturbed.[12] The United Kingdom instead adopted, well before generative AI, a deeming provision: Section 9(3) of the Copyright, Designs and Patents Act, 1988 provides that for a computer-generated literary, dramatic, musical or artistic work having no human author, the author is deemed to be "the person by whom the arrangements necessary for the creation of the work are undertaken" language strikingly similar to India's Section 2(d)(vi), and in fact one of the models India's own provision drew upon.[13] The only judicial interpretation of Section 9(3), Nova Productions Ltd v. Mazooma Games Ltd, held that the programmer who devised a videogame's underlying rules and logic, rather than the player whose input shaped each frame, had made the "necessary arrangements" an analogy of some relevance to whether an AI developer or an end user "causes" a generative-AI work to be created, though the case predates generative AI and concerned a far more deterministic system.[14] India therefore sits between two more doctrinally settled positions: the American refusal to protect AI output absent human authorship, and the British deeming fiction that has itself received only sparse judicial attention. Unlike either jurisdiction, India has neither a court ruling squarely on point nor a Copyright Office practice direction, leaving Section 2(d)(vi) formally broader than the U.S. rule but functionally less certain than the UK's.
V. CRITICAL ANALYSIS: A PROVISION WITHOUT INTERPRETIVE INFRASTRUCTURE
The central problem this analysis exposes is not that Indian law lacks a textual hook for AI-generated works Section 2(d)(vi) supplies one but that India lacks the interpretive infrastructure to apply it consistently. Three deficiencies stand out. First, the absence of a statutory or judicial definition of "computer-generated work" leaves the Copyright Office to decide, application by application and without published reasoning, whether a given AI-assisted work falls within the provision at all, as the RAGHAV episode illustrates.[15] Second, the Eastern Book Company originality standard was developed for human editorial labour and has never been tested against AI-generated content, so it is unclear whether prompting, parameter selection, or post-generation curation supplies the "skill and judgment" the Supreme Court required. Third, unlike the United Kingdom, where Nova Productions at least identifies a candidate test (who made the "arrangements necessary"), India has no comparable judicial guidance identifying whether the AI developer, the platform deployer, or the prompting user is "the person who causes the work to be created" where their contributions diverge, as they typically do in generative-AI use.[16] Existing Indian law is, on this account, formally more permissive than the U.S. rule but practically less usable, because commercial certainty depends on consistent application, which the Copyright Office's own conduct in the RAGHAV matter shows it has not yet achieved.
FINDINGS AND DISCUSSION
The doctrinal analysis yields four principal findings. First, Section 2(d)(vi) is textually capable of extending authorship to generative-AI output because it was drafted to cover "computer-generated" works without requiring direct human execution, but the provision has never been authoritatively construed by an Indian court in the generative-AI context, and its key phrase "the person who causes the work to be created" remains open to at least three competing readings (developer, deployer, or prompting user) (Research Question 1). Second, the Eastern Book Company "skill and judgment" standard is, in principle, well suited to assessing AI-assisted creativity because it already rejects both a purely mechanical "sweat of the brow" test and an unduly demanding novelty requirement, but it has not yet been applied to prompt-based or AI-curated works, leaving the originality threshold for such works undetermined (Research Question 2). Third, the comparative analysis shows that India's statutory language most closely resembles the United Kingdom's Section 9(3) deeming provision, while the United States has foreclosed protection for AI-authored works altogether following Thaler v. Perlmutter; India is textually closer to the UK model but lacks even the UK's limited judicial guidance from Nova Productions (Research Question 3). Fourth, because the uncertainty stems from an undefined statutory term and an untested judicial standard rather than from any inherent unsuitability of Section 2(d)(vi), both legislative clarification of "computer-generated work" and a definitive judicial ruling arising from the pending RAGHAV dispute would resolve the problem but legislative reform is preferable because it would apply prospectively and uniformly, whereas litigation would resolve only the specific facts before the court (Research Question 4).[17] Collectively, these findings indicate that India's copyright framework is not, in principle, inadequate to the challenge of generative AI, but its administration has been inconsistent, unreasoned, and commercially destabilising precisely where clarity is most needed.
CONCLUSION
The central legal issue examined in this article is whether Section 2(d)(vi) of the Copyright Act, 1957 a provision enacted in 1994 for a pre-generative-AI world can coherently accommodate the authorship questions raised by contemporary generative AI. The doctrinal analysis shows that the provision is textually broad enough to do so, distinguishing India from the United States' strict human-authorship rule confirmed in Thaler v. Perlmutter, and placing it closer to the United Kingdom's Section 9(3) deeming provision. Yet the Copyright Office's contradictory treatment of the RAGHAV registration, and the absence of any judicial application of the Eastern Book Company originality standard to AI-generated works, show that India has not translated this statutory head start into interpretive certainty. The existing legal framework is therefore best described as structurally adequate but administratively unreliable: the right statutory tool exists, but it has not been defined, tested, or consistently applied. Legislative clarification of "computer-generated work" and the identity of the "person who causes the work to be created," together with reasoned Copyright Office practice and a definitive judicial ruling in the RAGHAV dispute, would convert India's early but unstable statutory advantage into genuine legal certainty for creators and businesses relying on generative AI.
Reference
[1] "AI-Generated Content & Copyright Gap in Indian Law," Khurana & Khurana, 24 June 2026, www.khuranaandkhurana.com/ai-generated-content-and-the-copyright-act-1957-is-there-a-gap-in-indian-law.
[2] Copyright Act, 1957, No. 14 of 1957, Gazette of India, Sec. 2(d)(vi), as amended by the Copyright (Amendment) Act, 1994.
[3] "Exclusive: India Recognises AI as Co-Author of Copyrighted Artwork," Managing Intellectual Property, 9 June 2022, www.managingip.com/article/2a5bqo2drurt0bxl7ab24/exclusive-india-recognises-ai-as-co-author-of-copyrighted-artwork.
[4] "Recognition of AI Co-Author RAGHAV — An Anomaly?" Centre for Study and Research in Intellectual Property Rights, NUSRL, 11 March 2022, csriprnusrl.wordpress.com/2022/03/11/recognition-of-ai-co-author-raghav-an-anomaly/.
[5] Eastern Book Company v. D.B. Modak (2008) 1 SCC 1 (Supreme Court of India); Thaler v. Perlmutter, 130 F.4th 1039 (U.S. Court of Appeals for the D.C. Circuit, 18 March 2025); Nova Productions Ltd v. Mazooma Games Ltd [2006] EWHC 24 (Ch) (High Court of England and Wales).
[6] Copyright Act, 1957, Sec. 2(d)(vi).
[7] "AI-Generated Content & Copyright Gap in Indian Law," Khurana & Khurana, 24 June 2026.
[8] "Exclusive: India Recognises AI as Co-Author of Copyrighted Artwork," Managing Intellectual Property, 9 June 2022.
[9] "Recognition of AI Co-Author RAGHAV — An Anomaly?" Centre for Study and Research in Intellectual Property Rights, NUSRL, 11 March 2022.
[10] Eastern Book Company v. D.B. Modak (2008) 1 SCC 1.
[11] Ibid.
[12] Thaler v. Perlmutter, 130 F.4th 1039 (D.C. Cir. 2025); Baker Donelson, "Supreme Court Denies Certiorari in Thaler v. Perlmutter," 5 March 2026, www.bakerdonelson.com/supreme-court-denies-certiorari-in-thaler-v-perlmutter-ai-cannot-be-an-author-under-the-copyright-act.
[13] Copyright, Designs and Patents Act 1988 (UK), Sec. 9(3); "The UK's Curious Case of Copyright for AI-Generated Works: What Section 9(3) Means Today," Authors Alliance, 19 May 2025, www.authorsalliance.org/2025/05/19/the-uks-curious-case-of-copyright-for-ai-generated-works-what-section-93-means-today/.
[14] Nova Productions Ltd v. Mazooma Games Ltd [2006] EWHC 24 (Ch).
[15] "Recognition of AI Co-Author RAGHAV — An Anomaly?" Centre for Study and Research in Intellectual Property Rights, NUSRL, 11 March 2022.
[16] Nova Productions Ltd v. Mazooma Games Ltd [2006] EWHC 24 (Ch).
[17] Eastern Book Company v. D.B. Modak (2008) 1 SCC 1; Thaler v. Perlmutter, 130 F.4th 1039 (D.C. Cir. 2025).