I. Introduction

Imagine a generative AI system that, left to its own devices over a weekend, produces a full-length novel, and the developer submits it for copyright registration. The registrar, quite reasonably, asks: who is the author? The developer says: “the machine.” And the machine, of course, says nothing.

This is not speculative fiction. Stephen Thaler litigated exactly this before the US Court of Appeals for the Federal Circuit and the UK Supreme Court, arguing that his AI system DABUS deserved recognition as author and inventor. Both courts said no.[1] Neither answered the obvious follow-up: then who owns the output? India has not even reached the question.

India’s Copyright Act, 1957 was written when the most advanced creative tool was a photocopier. It has not been meaningfully amended since 2012. In an economy that NITI Aayog has identified as a priority destination for AI investment, this legislative silence is no longer a harmless quirk but a structural policy failure.

II. The ‘Human Author’ Problem

Copyright law everywhere begins with the same assumption: authors are human. In India, Section 2(d) of the Copyright Act defines ‘author’ with reference to persons like the author of a literary work, the composer of music, the artist of a painting. The underlying bargain is simple: creative labour deserves a temporary monopoly over reproduction and distribution.

Indian courts have elaborated this requirement. In Eastern Book Company v D B Modak (2008),[2] the Supreme Court held that copyright does not require mere labour or expense but a “modicum of creativity” which is the exercise of skill and judgment that is more than trivial, a formulation drawn from University of London Press v University Tutorial Press (1916)[3]. In RG Anand v Delux Films (1978), the Supreme Court reinforced that copyright subsists only in the expression of original thought from an identifiable human creator[4]. None of these formulations contemplate a non-human originator.

Cameras, word processors, and AutoCAD refined the mechanics of creation, but never altered its authorship: the human remained the source of creative judgment. The human’s intellectual contribution was always traceable. Generative AI is different in kind. When a user prompts an image generator with ‘paint me a Mughal miniature of a cyberpunk Delhi,’ the model’s internal processes make the substantive creative choices. The prompt is closer to a commission than a brushstroke. Applied to the Eastern Book Company standard, a minimal prompt provides neither the skill nor the judgment that Indian copyright doctrine demands of an author.

The US Copyright Office’s 2025 AI Copyrightability Report confirmed the logical consequence: purely AI-generated content without meaningful human creative control will not receive copyright protection. Without copyright, AI outputs enter the public domain instantly. The incentive structure for innovation and protection of human creators collapses.

III. India’s Accidental Advantage — and Its Accidental Gap

Section 2(d)(vi) of the Copyright Act defines the author of a “computer-generated work” as “the person who causes the work to be created.” This provision, borrowed in spirit from Section 9(3) of the UK’s Copyright, Designs and Patents Act 1988, gives India a statutory hook most jurisdictions lack: it could assign copyright in AI output to whoever deployed the system. However, Section 2(d)(vi) was never crafted with generative AI in mind. Its conceptual foundation lies in a model of software that yields structured, foreseeable outputs, such as payroll reports or pre-formatted design templates. Extending it to a model capable of autonomously generating a 90,000-word novel exposes clear interpretive gaps.

What does it mean to “cause” a work when the creative choices are made by the model itself? In Camlin Pvt Ltd v National Pencil Industries (1985),[5] the Delhi High Court held that direction and control over the creative process are relevant indicators of “causing” a work. On that standard, a user who types a three-word prompt and receives a complete novel has likely not “caused” the work in any doctrinally meaningful sense. No Indian court has answered this question. The National IPR Policy 2016 speaks of adapting IP law to new technologies without mentioning AI; the NITI Aayog’s National AI Strategy (2018) focuses on economic opportunity without touching IP. The result is a framework with the statutory vocabulary for AI authorship but no case law, policy guidance, or reform process to deploy it.

IV. The Global Conversation India Is Missing

Other jurisdictions are not standing still. WIPO’s Revised Issues Paper on IP and AI (2020) identified seven policy options for AI authorship and called on member states to develop considered national positions. India has submitted none.

The EU Artificial Intelligence Act (2024)[6] introduces transparency and disclosure obligations for AI-generated content, leaving ownership to member-state copyright law. Treating disclosure as a regulatory obligation and ownership as a copyright issue is a structural distinction worth adopting. India, however, should not simply transplant the EU model. Its creative economy is characterised by fragmented rights, informal licensing, and a vast unregistered creative sector; an opt-out disclosure regime of the EU type may impose compliance costs on small creators that dwarf any benefit.

The United Kingdom’s Section 9(3), which directly inspired Section 2(d)(vi), has faced instructive interpretive pressure. In Nova Productions Ltd v Mazooma Games Ltd (2007), the Court of Appeal held that the relevant “author” was the programmer whose skill shaped the generative process, not the end-user who merely triggered it[7]. If Indian courts adopt a similar reading, the developer or deployer, not the prompt-author would be the presumptive owner of AI outputs. The implications for enterprise-deployed AI are significant.

The ongoing New York Times v OpenAI litigation has forced the training-data question into the open: does scraping copyrighted works to build AI systems constitute infringement?[8] Under India’s fair dealing provisions in Section 52 of the Copyright Act, the answer is genuinely unclear. India currently lacks a text-and-data-mining exception comparable to EU Directive 2019/790 or Japan’s Article 30-4 of its Copyright Act. Every Indian AI developer training on Indian-origin content therefore operates in a zone of legal uncertainty that chills both investment and legitimate research.

India has distinctive stakes. Given that its technology sector includes some of the world’s largest AI service providers, it has a direct interest in how AI outputs are protected. Its creative industries, Bollywood, a thriving music ecosystem, and one of the world’s largest publishing markets, have an equally direct interest in ensuring AI does not hollow out the economic value of human creativity. A considered legislative framework would need to balance both. There is currently no framework at all.

V. Conclusion: Towards a Principled Framework

Three targeted interventions could resolve the most pressing ambiguities without wholesale revision of the Copyright Act.

First, Parliament should clarify Section 2(d)(vi) by defining what it means to “cause” a generative AI work, with ownership vesting in the entity that exercises sufficient “purposive direction” over the output. Drawing on Eastern Book Company and Nova Productions, the following non-exhaustive factors would operationalise the standard: (i) the specificity of creative instruction in the prompt or model configuration; (ii) iterative human involvement in selecting and curating outputs; (iii) economic risk-bearing by the entity that bears the commercial consequences of the output’s quality; and (iv) degree of control over the model’s training and architecture. On this framework, a developer who fine-tunes a model on proprietary data and deploys it in a constrained professional context would likely qualify; a consumer who types a brief prompt and accepts the first output would likely not. Minimal prompts would presumptively yield public-domain outputs.

Second, a disclosure obligation: any work produced substantially by AI should carry a machine-readable label identifying it as such, implemented through an amendment to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules 2021 rather than as a copyright condition, preserving compliance symmetry for small creators.

Third, the Law Commission should examine whether India’s fair dealing provisions extend to computational analysis, with a focused India-specific answer on training data, specifically whether a narrowly tailored text-and-data-mining exception would serve India’s research interests, and whether an opt-out framework would adequately protect Indian rights-holders.

None of this requires Parliament to decide whether AI is a person. That metaphysical question can wait. What cannot wait is the practical one: India has a booming AI economy, a rich creative sector, and a 1957 Copyright Act that has no idea either of them exists. “Creation no longer waits for a human hand, but the law still does.”

[1]Thaler v Vidal 21-2347 (Fed Cir, 5 August 2022); Thaler v Comptroller-General of Patents [2023] UKSC 49.

[2]Eastern Book Company v D B Modak (2008) 1 SCC 1 [15]–[17].

[3] University of London Press Ltd v University Tutorial Press Ltd [1916] 2 Ch 601.

[4] RG Anand v Delux Films AIR 1978 SC 1613, [46]–[48].

[5]Camlin Pvt Ltd v National Pencil Industries AIR 1986 Del 444.

[6]Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act) [2024] OJ L 2024/1689, arts 50–53.

[7] Nova Productions Ltd v Mazooma Games Ltd [2007] EWCA Civ 219 [105]–[106].

[8] The New York Times Company v Microsoft Corporation and OpenAI Inc 1:23-cv-11195 (SDNY, filed 27 December 2023).

Aryan Gupte is a fourth-year law student at Jindal Global Law School. His academic interests include Intellectual Property Rights, international arbitration, and AI governance.

Shreemayi Pathak is a fourth-year law student at Jindal Global Law School. Her academic interests lie primarily in Intellectual Property Rights, with a particular focus on fashion and luxury law, as well as media law.