Artificial Intelligence and Intellectual Property Rights in India: Who Owns AI-Generated Content?: Author: VAIBHAV SINHA
Generative AI appears to be threatening one of the most fundamental premises of copyright law, i.e., the idea that there must be a legally recognised author whose expression of creativity can be attributed to him. This does not mean that Indian law offers no starting point. Section 2(d)(vi) of the Copyright Act, 1957 identifies the author of a computer-generated work as “the person who causes the work to be created”.
ARTICLE
Abstract
Generative AI appears to be threatening one of the most fundamental premises of copyright law, i.e., the idea that there must be a legally recognised author whose expression of creativity can be attributed to him. This does not mean that Indian law offers no starting point. Section 2(d)(vi) of the Copyright Act, 1957 identifies the author of a computer-generated work as “the person who causes the work to be created”. The difficulty is applying that rule when creation involves training data, model architecture, prompts and automated generation. This article attempts to answer this question from two different angles, that is, input rights [i.e., the copyrighted works used to train the AI systems] and output rights [i.e., authorship and ownership of the works produced by an AI system]. Recent developments in India suggest that while the foundational concepts are relevant, the contours of those concepts are becoming increasingly obscure.
Keywords
Artificial Intelligence, Copyright, Generative AI, Authorship, Ownership, Training Data, India
Introduction
Generative AI creates an unusual copyright problem because the person operating a system may not determine the precise expression that emerges from it. Yet the ownership question cannot be considered in isolation. Indeed, the question of ownership of the output of AI systems can only follow from the very preliminary question of whether the reproduction of the copyrighted works that went into the training of these AI systems was authorised at all.
The two questions should therefore be separated. Input rights concern training material while output rights concern the resulting work.
I. The Statutory Starting Point
The Copyright Act already contains language relevant to computer-generated works. Section 2(d)(vi) defines their author as “the person who causes the work to be created.” Section 13 addresses copyright in original works Section 14 sets out exclusive rights and Section 17 generally makes the author the first owner subject to statutory exceptions.
The Act therefore does not require a computer to become an author. Instead it requires identification of a legally recognised person responsible for creation. Generative AI makes that inquiry difficult because several actors may be involved: the developer, system operator, prompt writer, or user who selects and edits the output.
Authorship must also be distinguished from originality. In Eastern Book Company v. D.B. Modak, the Supreme Court recognised that copyright requires a minimum degree of creativity. A work may therefore satisfy the originality threshold without automatically resolving who its legal author
II. ANI v. OpenAI: Copyright at the Training Stage
The input-side dispute is illustrated by ANI Media Pvt. Ltd. v. OpenAI OpCo. ANI alleged unauthorised copying and storage of its copyrighted material for training large language models and alleged that AI outputs could reproduce protected expression.
The Delhi High Court treated storage of copyrighted works for training as reproduction but at the interim stage considered OpenAI's reliance on Section 52(1)(a), including its reference to “private or personal use, including research”. The Court examined whether use was confined to training or whether it prejudiced the copyright owner's legitimate commercial interests or caused market substitution and whether it advanced public interest.
The Court considered “research” broad enough prima facie to encompass machine learning in the circumstances before it. It also found that the material presented did not establish memorisation or substantial reproduction of ANI's protected expression through the outputs examined.
The decision cannot be relied upon as an authoritative statement of the copyright position vis-à-vis AI training. It concerned interim relief and expressly preserved the distinction between prima facie findings and final adjudication. The more important aspect of this order is how the existing principles of fair dealing are being stretched to apply to technology that was not even envisaged by the legislature.
III. A Recent Entrance to Paradise: Originality Without AI Authorship
The output-side issue appears in the Copyright Office proceedings concerning Dr Stephen L. Thaler's A Recent Entrance to Paradise. The Office found that the work satisfied the limited originality threshold under Section 13, but refused to recognise DABUS as its author under Section 2(d)(vi).
The reasoning separates originality from authorship. DABUS was not a natural or juristic person recognised in law and therefore could not occupy the statutory position of author. The Office instead examined the broader causal process, including human configuration, inputs and initiation, and focused on identifying the legally recognised person who materially caused the particular work to be created.
The proceedings also show why ownership of an AI system is insufficient by itself. The Office noted that owning DABUS did not create a general entitlement to copyright in everything generated through it.
IV. From Interpretation to Reform
Between two extremes lies a continuum of AI-created works. At one end are AI-assisted work, where human choices substantially determine expression. The middle consists of human-directed generation involving meaningful prompting, selection, editing or curation. At the other end are highly autonomous outputs where meaningful human creative intervention becomes difficult to identify.
Section 2(d)(vi) can accommodate substantial human involvement, but its application becomes less certain as human contribution becomes more remote.
The training-data problem creates a parallel challenge. DPIIT's December 2025 Working Paper on Generative AI and Copyright considers licensing and remuneration mechanisms intended to balance copyright interests with AI development. These remain policy proposals not enacted law.
Future reform could clarify the meaning of “person who causes the work to be created”, improve transparency concerning training datasets, and consider licensing or collective licensing mechanisms. Such reform should remain technology-neutral and avoid assuming either that every AI output deserves copyright or that every use of copyrighted material for training is necessarily infringement.
Conclusion
The question “who owns AI-generated content?” has no single answer under India's present framework. Input rights concern the lawful use of copyrighted works for training; output rights concern originality, authorship and ownership of the resulting expression.
ANI demonstrates judicial adaptation of existing reproduction and fair-dealing principles while A Recent Entrance to Paradise demonstrates that originality does not automatically make an AI system a legal author. Section 2(d)(vi) provides a useful bridge between computer-generated expression and human authorship, but increasing AI autonomy exposes the limits of that formulation.
The emerging challenge is therefore one of legal clarification: protecting creators and preserving copyright incentives while leaving sufficient space for technological development, research and public interest.
References / Citations
Copyright,1957, S- 2(d)(vi), 13, 14, 17, 52. [Copyright Act, 1957]
ANI Media Pvt. Ltd. v. OpenAI OpCo LLC, 2026 SCC OnLine Del 5291 (Delhi High Court, 24 July 2026). [SSCOnline]
ANI Media Pvt. Ltd. v. OpenAI OpCo LLC, 2026 SCC OnLine Del 5478 (Delhi High Court, 27 July 2026). [SSCOnline]
Eastern Book Company v. D.B. Modak, (2008) 1 SCC 1. [SSCOnline]
Copyright Office, Dr Stephen L. Thaler — “A Recent Entrance to Paradise”, Diary No. 9356/2022-CO/A. [PDF]
Department for Promotion of Industry and Internal Trade, Government of India, Working Paper on Generative AI and Copyright, Part I: One Nation One License One Payment—Balancing AI Innovation and Copyright (December 2025). [PDF]
Avishek Chakraborty, “Authorship of AI Generated Works under the Copyright Act, 1957: An Analytical Study”, Nirma University Law Journal. [SSRN.COM]
Saakshi Agarwal & Chintan Bhardwaj, “The Dilemma of Copyright Law and Artificial Intelligence in India”. [SSRN.COM]