Published On: March 15th 2026
Authored By: Ankita Ghosh
Brainware University
Abstract
Artificial intelligence has quietly infiltrated the domain of creative expression, producing poems, paintings, and legal documents with minimal human input. This development poses fundamental challenges to copyright law, which was built on the assumption of human authorship. This article examines the intersection of AI-generated content and copyright under Indian law, with reference to comparative jurisprudence. It addresses the questions of authorship, ownership, originality, infringement through training data, and moral rights, and argues that existing frameworks, while not entirely irrelevant, are insufficient to address the novel realities of generative AI.
I. Introduction
A novel writes itself.
A song composes without a composer.
A picture appears without an artist.
Sounds impossible. Until it was artificially done.
Artificial Intelligence (AI) has quietly made its way into the sphere of creativity: it writes poems, creates logos, and drafts legal documents. What once required human imagination now requires only a prompt and a click. Efficient. Powerful. Unsettling.
And legally confusing.
Copyright law, in India as elsewhere, was founded on the assumption that creativity is a human endeavour. AI, however, does not reason like a human. It learns. It predicts. It generates. Between code and creativity, there exists a legal vacuum. A grey area.
This article examines AI-generated content and copyright through the lens of legal principles, statutory interpretation, and emerging Indian and foreign case law. It addresses urgent questions of authorship, ownership, originality, and infringement.
II. AI-Generated Content: The Basics
AI-generated content refers to content produced by a machine learning model, specifically Generative AI systems such as ChatGPT and Midjourney. These systems are trained on vast quantities of data including books, images, music and articles, much of which is protected by copyright.
Here lies the first tension.
AI does not imitate in the classical sense.
But neither does it create per se.
It generates outputs through pattern recognition derived from existing works. Is this inspiration? Or reproduction?
Copyright law was not designed to answer that question.
III. Indian Copyright Law: The Human Requirement of Authorship
Indian copyright jurisprudence is anchored in the Copyright Act, 1957. Section 13 protects original works of a literary, dramatic, musical, or artistic nature. The Act, however, makes no mention of non-human creators.
Authorship Under Indian Law
Section 2(d) defines “author” as follows:
In relation to a literary or artistic work: the person who makes the work.
In relation to a computer-generated work: the person who causes the work to be created.
This provision appears promising. But ambiguity persists.
Does “person” include AI?
Or only humans who use AI?
Indian courts have consistently emphasised human creativity. In Eastern Book Company v. D.B. Modak,[1] the Supreme Court held that originality requires a modicum of creativity, not mere labour. Such creativity presupposes human intellectual effort.
AI has no intellect. No intention. No consciousness.
Under Indian law, therefore, AI cannot be an author.
IV. The Ownership Question: Who Owns AI-Generated Works?
Accepting that AI cannot be an author, the question becomes: who owns the output?
Possible claimants include:
The programmer
The user
The AI developer company
No one (public domain)
Indian law addresses this indirectly through Section 2(d)(vi), which provides that for computer-generated works, the author is the person who causes the work to be created.
This raises a further question.
Is typing a prompt sufficient?
Comparative Case Law
In Narendra Publishing House v. Rajiv Beri,[2] the Delhi High Court emphasised that craft and judgment in the act of creation are essential. Mere facilitation does not confer ownership.
Applying this reasoning, a user who types a simple prompt (“write a poem about rain”) may not exercise sufficient creativity to claim authorship of the result.
But what of complex prompts? Iterative refinement? Substantial human intervention in shaping the output?
The legislation provides no bright-line rule. Uncertainty prevails.
V. Originality and AI: Can Machine-Generated Content Be Original?
Copyright protection turns on originality.
In University of London Press Ltd. v. University Tutorial Press Ltd.,[3] the court held that originality means the work originated from the author, not that it was novel. This standard was adopted by Indian courts in Eastern Book Company v. D.B. Modak.[4]
AI output does not emerge from a conscious mind. It is a product of statistically probable patterns derived from training data.
Therefore:
The expression may appear new.
The process is, nonetheless, derivative.
This creates a paradox: a work may appear original while lacking the legally recognised quality of originality.
Some scholars argue that AI output should belong to the public domain. Others caution that this would discourage investment in creative tools. There is no consensus.
VI. Copyright Infringement: The Training Data Dilemma
This is perhaps the most contested issue.
AI systems are trained on large datasets, typically scraped from the internet. Books. Paintings. Songs. Articles. A significant proportion of this material is copyrighted.
Is training AI on copyrighted data an act of infringement?
Section 52 of the Copyright Act, 1957 recognises fair dealing exceptions for research, private use, criticism, and review. However, AI training is commercial in purpose, automated in execution, and industrial in scale. It is difficult to characterise such use as “fair.”
Global Case Law Influence
India has not yet produced direct case law on this issue, but foreign decisions offer useful guidance.
In Authors Guild v. Google Inc.,[5] the United States Court of Appeals for the Second Circuit upheld the digitisation of books as fair use on the basis that the use was transformative. AI developers have relied on this reasoning to defend their training practices.
However, in Getty Images v. Stability AI Ltd.,[6] Getty Images alleged that its copyrighted photographs were used to train an AI model without permission. The case remains pending, but it signals judicial discomfort with unregulated data scraping.
Indian courts may adopt a stricter approach, particularly in light of the moral rights provisions under Section 57 of the Copyright Act.
VII. Moral Rights and AI
Moral rights protect an author’s:
Right of attribution
Right to integrity of the work
AI threatens both.
A generative AI system can produce work in the recognisable style of a living artist, without attribution and without consent.
In Amarnath Sehgal v. Union of India,[7] the Delhi High Court affirmed the enduring significance of moral rights even after an assignment. The court’s reasoning suggests that any act which denigrates an artist’s reputation or distorts their work may constitute an infringement of moral rights.
When AI replicates an artist’s style and undermines their livelihood, the question of moral rights infringement becomes not merely possible but pressing.
VIII. The Story of an Artist and an Algorithm
Consider this scenario.
A digital artist in Mumbai spends many years perfecting her style. Her portfolio is publicly accessible online. An AI model is trained on her work. Before long, clients begin using AI to generate images in her distinctive style, at no cost.
Her work is not copied.
Her signature is not forged.
But her livelihood disappears.
Her protection under copyright law is an uphill battle.
This is not a hypothetical. It is already happening.
IX. Policy Gaps and the Need for Reform
India does not yet have a dedicated legal framework governing AI and intellectual property.
NITI Aayog’s National Strategy on Artificial Intelligence (2018) does address ethical considerations relating to AI, but it does not provide legal remedies for copyright-related harms.
Possible reforms include:
Mandatory disclosure of training datasets.
Licensing regimes for the use of copyrighted data in AI training.
Recognition and crediting of AI-assisted works (as distinct from fully AI-generated works).
Clear rules for attribution and ownership of AI-generated output.
The law must evolve. Thoughtfully. Carefully.
X. Conclusion
AI does not violate copyright law in isolation. It is the manner in which AI is developed, trained, and deployed that creates the conditions for infringement. The legal framework, developed to govern quills and printing presses, now confronts algorithms and neural networks. Indian courts have long upheld human creativity, moral rights, and originality as foundational values. These principles retain their relevance; they are, however, no longer sufficient on their own.
The grey area cannot resolve itself.
It requires judicial interpretation, legislative intervention, and ethical restraint.
Until that time, AI-generated content will remain in legal limbo: potent, commercially valuable, and deeply unsettling.
The tension between what AI can do and what the law permits will shape the future of creativity.
Between code and conscience.
Between machine and mind.
References
[1] Eastern Book Company v. D.B. Modak, (2008) 1 SCC 1 (India).
[2] Narendra Publishing House v. Rajiv Beri, (2009) 39 PTC 358 (Del) (India).
[3] University of London Press Ltd. v. University Tutorial Press Ltd., (1916) 2 Ch 601 (UK).
[4] Eastern Book Company v. D.B. Modak, (2008) 1 SCC 1 (India).
[5] Authors Guild v. Google Inc., 804 F.3d 202 (2d Cir. 2015) (USA).
[6] Getty Images v. Stability AI Ltd., UK High Court, 2023 (pending, status to be verified).
[7] Amarnath Sehgal v. Union of India, (2005) 30 PTC 253 (Del) (India).
