Generative AI And Copyright Under Indian Law

Author: LawSphereX

Category: Intellectual Property

Published: 2026-08-23T19:57:46.390Z

Updated: 2026-08-30T07:11:52.460Z

GENERATIVE AI AND COPYRIGHT UNDER INDIAN LAW

Jayant Bhardwaj, OP Jindal Global University

ABSTRACT

The rise of artificial intelligence (AI) has called into question some of copyright law's most established presumptions. When machines write, paint, or produce music, the question of ownership of the works of art arises unavoidably. This study explores this topic using Indian copyright law as a paradigm. It examines the fundamental principles of the Copyright Act of 1957, namely originality, authorship, fixation, and fair dealing, and compares them to recent global legal cases like Authors Guild v. Google Inc. (2015) and Andy Warhol Foundation v. Goldsmith (2023). The study contends that, while the Indian copyright system continues to rely on human invention, it may change in tandem with technology innovations due to its malleable legislative language and public-interest-oriented jurisprudence.

The study provides an interpretive framework that preserves human responsibility, acknowledges AI-supported creativity, and upholds the basic balance achieved in intellectual property law between access and protection.

INTRODUCTION

According to classical understanding of copyright law, author is a talented, intellectual, and creative individual. Every part of the 1Copyright Act of 1957 that mentions "author," "owner," or "moral right" paints a picture like this. Nonetheless, the support it received from the human creative notion is now emphasized. This debate focuses on generative AI systems like ChatGPT, DALL·E, and Midjourney, which produce expressive works like human art with minimal human input. They create such works by leveraging the vast accumulated knowledge of prior works; consequently, creation is neither mechanical nor repetitive. The major issue is that, AI systems employ statistics to create a finished output, whereas copyright law relies on humans' mental efforts to safeguard the result. What if people no longer saw creativity and originality as human traits?

How would the law define and enforce accountability? The core of this legal issue is whether copyright should protect the value of the finished work or the time and effort put into its creation. AI research may come to a stop due to a theological issue, but it also has a chance in India, where the state's copyright restrictions are sometimes ambiguous and hence encourage competing viewpoints. As a result, one of the most difficult tasks is determining how to maintain copyright current while protecting its fundamental ideals. MAIN BODY THE HUMAN BASIS OF COPYRIGHT The three factors numbered one through three have always been somewhat overlapping utilitarianism, personality, and labour all of which, to some extent, emphasize the same point: the basis of creativity is human activity.

To put it simply, 2John Locke's labour theory holds that a person has a claim to the property he or she has worked on, which is the product of combining resources and effort. Hegel's concept of personality regards creation as a reflection of one's moral and intellectual qualities. According to Bentham's utilitarian viewpoint, copyright serves as an incentive for people to create socially valuable works. 1 Ayush Verma, “Copyright Act, 1957” (iPleaders, March 30, 2020) accessed October 10, 2025. 2 Jason Rantanen, “William Fisher, Theories of Intellectual Property” (University of Iowa College of Law, December 3, 2024) accessed October 10, 2025.

AI-generated outputs cannot readily be classified into any of the existing models. Machines have no moral rights, objectives, or human-like traits. Their "labour" consists solely of running algorithms rather than intentional action. However, if the utilitarian argument makes it easier to reap advantages or broadens coverage, it may lead to the argument that a minimal level of protection should be provided. Thus, the entire dispute is about the law determining who the human author is the person who educated or controlled the AI rather than whether AI can copyright its creations. The term "person" under 3Section 2(d) of the Copyright Act is always synonymous with "author," but other names are used depending on the nature of the work. Section 13(1) requires that a creator be recognized, and copyright is limited to "original literary, dramatic, musical, and artistic works.

" As a result, Indian law specifies who can write. However, the definition of "computer-generated works" in Section 2(d)(vi), which defines the author as "the person who causes the work to be created," establishes the framework for the investigation of AI-produced content. The law does not require a tangible manifestation of the work, and causal agency may suffice. The often-overlooked characteristic becomes the critical point from which current interpretation is drawn. It avoids anachronism while maintaining the relationship between creativity and human control by permitting AI-assisted authorship in situations where human participation is important. CREATIVITY AND ORIGINALITY The originality doctrine is one of the standards used to determine protection. In 4Feist Publications v.

Rural Telephone Service (1991), the United States Supreme Court rejected the "sweat of the brow" requirement, instead requiring a "modicum of creativity. " In the 2008 case 5Eastern Book Company v. D. B. Modak, the Indian judiciary agreed that originality needed "independent skill and judgment," indicating some amount of intellectual exertion. The primary question remains whether AI-created art contains an intellectual component. A

3 accessed October 10, 2025.

4 “Feist Publications, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340 (1991)” (Justia Law)

accessed October 10, 2025.

5 “Eastern Book Company & Ors. v. D.B Modak & Ors., Delhi High Court, Judgment, Law, Casemine.Com”

(https://www. casemine. com) accessed October 10, 2025. photographer, for example, creates the composition using his talent and judgment, whereas a human who understands the process communicates explicit instructions or curates the end results. If the end output is classified as AI-assisted, it might be argued that Modak's inspiration for creation is met in these circumstances. The work does not meet the originality requirement if it is generated mechanically without any assistance from humans or control. Indian law protects human innovation, not algorithmic predictability. As a result, the evaluation now focuses on the level of human creativity rather than the nature of the technology. Once AI generates an output, the digital duplicate of the output is saved, making it very easy to encounter fixation the incarnation of an idea in a tangible medium.

The key issue, however, is the clash between idea and expression, as described in 6Baker v. Selden (1879) and later reaffirmed in the Indian case of R. G. Anand v. Deluxe Films (1978). The reason for this is that AI algorithms learn from a vast quantity of existing creativity, therefore the creator is merging different styles and developing new ones, and in the process causing uncertainty in this division. A thorough similarity analysis is then required to determine the case's infringement. Even if there is no direct copying, an AI output may violate copyright if it captures the expressive core of the training material. This problem is exacerbated when copyrighted items are included in datasets without consent which is frequent practice when training huge language models.

OWNERSHIP AND AUTHORSHIP In copyright, authorship serves two purposes: it assigns responsibility and identifies the rights owner. The 1989 United States case 7Community for Creative Nonviolence v. Reid defined authorship as having control over the creative process. The Indian Act section 2(d)(vi), which stipulates that the person who "causes" the job to be made, is in accordance with these functional conditions. After completing this task, a few plausible writers can be identified: 6 “Baker v. Selden, 101 U. S. 99 (1879)” (Justia Law) accessed October 10, 2025.

7 “Community for Creative Non-Violence v. Reid, 490 U.S. 730 (1989)” (Justia Law)

accessed October 10, 2025.

1. Coders who create algorithms.

2. The trainer oversees supplying data and getting it ready.

3. The user selects the output and inputs prompts.

4. The corporation or group that launches and sells the AI system.

According to Indian copyright law, authorship can be assigned to the user or developer who has the most factual influence over the creation process, based on the facts. This perspective is supported by Indian courts' proclivity for pragmatic, case-by-case reasoning. It also corresponds to Section 17's work-for-hire presumption, which states that ownership can be transferred to the employer or commissioner unless a contrary agreement exists. Instead of the theoretical dead end of computer legal personhood, this approach maintains human authorship and accountability. INNOVATION AND VIOLATIONS Making copies of large datasets is part of the AI system's training procedure. Under current Indian law (Section 14), such copying is considered reproduction, unless Section 52(1) provides an exception. The three most important exclusions are "private study," "research," and "use for education.

" However, these were meant for human examination rather than the computerized statistical analysis that was performed. The High Court of Delhi's decision in 8Rameshwari Photocopy Services (2016) legally enlarged the idea of fair use by permitting class photocopying, which provided a social and educational value at the time. Artificial intelligence training might be permissible if it served a similar public purpose, such as language preservation or scientific study. However, courts may conclude that there has been infringement if the training is for profit, as is the case with major private AI corporations, especially if the outputs are comparable to the original works. 9Civic Chandran v. Ammini Amma (1996) PTC 16 (Ker) continues to consider

8 Shreya Sampathkumar, “University of Oxford v. Rameshwari Photocopy Service” (IP Matters, March 30,

2022) accessed October 10, 2025.

9 “Civic Chandran & Others v. C. Ammini Amma & Others, Kerala High Court, Judgment, Law,

Casemine. Com” (https://www. casemine. com) accessed October 10, 2025. the principle of fair use, which is determined by the purpose, amount used, and market effect. Since 10Campbell v. Acuff-Rose Music (1994), the transformative use doctrine has been a key component of US law; it considers whether the new work contributes any new expression or meaning to the old one. The Supreme Court, in 11Andy Warhol Foundation v. Goldsmith (2023), narrowed this requirement by declaring that no commercial licensing could reduce similar expressive works to simple changes. Indian courts did not rigorously apply the transformational criteria, but nonetheless followed the same purposive analysis. The 2005 verdicts in Rameshwari Photocopy and Amar Nath Seghal v. Union of India demonstrate judges' willingness to mediate between moral and economic interests and the right to artistic expression.

The question in artificial intelligence is whether the outcomes of machine learning can be called transformative. An AI-generated new work that recontextualizes current expression, such as converting text into data visualization, could be considered a genuine variant. However, many models that just emulate style or structure run the risk of being labelled as derivative imitation rather than revolutionary invention. ACCOUNTABILITY AND MORAL RIGHTS Section 57 grants authors the right to assert authorship and prevent their work from being twisted under the moral rights doctrine. These rights were deeply personal and were forcefully reaffirmed in 12Amar Nath Sehgal (2005), when the Delhi High Court recognized their inalienable nature. AI is not a human being; hence it lacks moral rights. Human features of machine functioning may claim moral rights.

For example, even if the maker employs AI to produce and then refine visual notions, he remains the legal and moral creator of the final piece. The position of responsibility becomes more complicated when artificial intelligence violates

10 “Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569 (1994)” (Justia Law)

accessed October 10, 2025.

11 “Andy Warhol Foundation for Visual Arts, Inc. v. Goldsmith, 598 U.S. ___ (2023)” (Justia Law)

accessed October 10, 2025.

12 Janhavi KM, “Amarnath Sehgal v. Union of India” (IP Matters, August 22, 2022)

accessed October 10, 2025. the law. Sections 51 and 55 define infringement as the three acts of creating, reproducing, or disseminating the work to the public. This indicates that the user, developer, or corporation installing AI is liable, not the AI itself. Thus, the legal system that holds solely natural or juristic persons accountable maintains the human-centric accountability concept. GLOBAL VIEWPOINTS AND INDIA’S PATH The 13UK Copyright, Designs, and Patents Act 1988 specifically recognize computer-generated works by assigning authorship to the "person by whom the arrangements necessary for the creation of the work are undertaken. " This practical norm is analogous to Section 2(d)(vi) in India. In contrast, the US Copyright Office refuses to patent exclusively AI-generated works, such as Zarya of the Dawn (2023), and insists on human authorship. India is amid these two paradigms.

It blends the judicial tradition of striking a balance between individual rights and the public interest with statutory flexibility. It shown in the EBC vs. Modak and Rameshwari Photocopy cases that it is ready to interpret copyright in ways that promote public access, creativity, and education. Because of this perspective, India can develop a context-specific model for AI law that is neither entirely open nor strictly proprietary. Furthermore, as a signatory of the 14TRIPS Agreement and the Berne Convention, India is required to guarantee minimum protection requirements but may interpret these responsibilities considering national interests such as equal access to information and digital inclusion.

POLICY REFLECTIONS The relationship between AI and copyright involves not only technical copyright legal concerns, but also broad policy issues about what society wishes to reward and protect. Giving too much protection to AI outputs risks granting tech companies de facto monopolies over both data and algorithms. Simultaneously, no protection may inhibit human-AI collaboration. In cases such as 15Novartis v. Union of India (2013) and 16Bayer v. Union of India (2013),

13 “Copyright, Designs and Patents Act 1988”

accessed October 10, 2025. 14 “Role of the TRIPS Agreement and Its Scope in India” accessed October 10, 2025.

15 Guest Post, “Novartis A.G. v. Union of India

Case Analysis” (iPleaders, February 26, 2024) accessed October 10, 2025. 16 “Bayer Corporation vs Union of India” (LawFoyer | A daily doze for inquisitors, February 27, 2025) Indian copyright law has not only favoured the public interest but also emphasized technological accessibility. The same principle should be used to the works of AI developers. Copyright should ensure that the public has access to information, culture, and data while also allowing it to innovate. Lawrence Lessig argues in 17Code: Version 2.0 that it is the government's responsibility to prevent "code from replacing law" in situations when private code controls speech. To put it another way, Indian courts must avoid treating copyright as a monopoly over data and algorithms. What they should do is protect human originality while also providing cultural and educational resources for education and creativity.

CONCLUSION

India can manage the matter by judicial interpretation and precise policy modifications rather than a radical legislative reform:

1. Under Section 2(d)(vi), grant authorship to a human being helped by AI, with the person

coordinating as the author rather than the AI.

2. Include non-expressive artificial intelligence for innovation and research inside the fair

dealing clause of Section 52.

3. In accordance with Section 13, strictly enforce originality standards to prohibit the

acquisition of rights in simply mechanical outputs.

4. To ensure accountability, Sections 51 and 55 shift liability to human actors.

5. Section 57 protects the moral rights of the designated human donor.

6. Facilitate the establishment of open datasets for non-profit AI research, in line with

India's right to information. These moves will allow the Indian copyright system to evolve naturally while adhering to its commitments to other countries and maintaining doctrinal coherence, while also promoting accessed October 10, 2025.

17 LESSIG, “Codev2” (LESSIG) accessed October 10, 2025.

innovation and equity. Generative AI has challenged the definition of creativity, but not its goal. The copyright goal has always been the same: to promote innovation for the benefit of society. Even though machines can help with the creative process, they cannot be considered intellectual or moral creators. In the legal environment, there should be a clear distinction between AI as a creator and AI as a tool. Indian copyright law, which is founded on fairness and flexibility, already has the interpretive tools needed to deal with this development. Through a well-thought-out and public-spirited interpretation of the current provisions Sections 2(d)(vi), 13, 14, 52, and 57 India can develop a comprehensive strategy that promotes technical advancement while rewarding human ingenuity.

The key question in the long run is not whether AI can "create" art, but whether the law will adapt to a multifaceted, collaborative, and yet-to-be creative environment. The Indian system is an excellent fit for dealing with this issue since it combines court practicality with legal simplicity, ensuring that copyright is always up to date and not just a remnant from pre-digital times.