A Legal Analysis of Generative AI vs. Ownership.

Author: Reeshav Das
Student, LL.M.
(IMS UNISON UNIVERSITY,
Dehradun, Uttarakhand)
Batch: 2025-26

Co-Author: Rahul Singh
Assistant Professor School of law
(IMS Unison University Dehradun)

Keywords: Generative AI, copyright ownership, artificial intelligence, authorship, Indian copyright act, machine autonomy, IP law, digital creativity, human supervision, AI generated works.

Abstract:

Generative artificial intelligence (AI) has presented significant challenges to the doctrine of copyright, especially when it comes to determining authorship, ownership and originality in automated creative processes. As tools like ChatGPT, Midjourney, DALLE, and music generation models enter the artistic, literary and professional workflow, traditional copyright principles, which are based on human creativity, are faced with unprecedented tension.By relating to the nature of the generative AI systems, the legal status of the AI-generated content according to the Indian law, the existing doctrinal dilemmas in the context of authorship and creativity, and comparative approaches across different countries, the paper surpasses that there is a need to change the Indian legal framework that is highly ambiguous in the relation to the ownership of the works created by AI to be sure that such innovation, human agency and the interests of people are not harmed. The study ends with recommendations for statutory intervention, interpretation of judiciary and regulation guidelines that can effectively balance technology advancement and copyright principles.6

Introduction

Generative AI models have completely altered the modern conception of creativity, authorship, and the nature of expressive works. These systems, which have been trained on vast amounts of data, can write poems, and illustrations, compose software code, create musical compositions, architectural designs, and even legal documents that are entirely their own creation in a wholly autonomous way the growing capabilities of these systems challenge the classical underpinnings of copyright law, which has always focused on human authorship and the exercise of independent skill and judgment. Across jurisdictions throughout the world, courts and policy makers are struggling to answer questions including: Can a machine be an author? Who owns AI-generated works? Does human supervision make up for ownership?What occurs when AI is creating without the human input that can be predicted?These questions are still very much outstanding in India, especially with the growing integration of generative AI in creative and commercial processes by industries. The Indian Copyright Act, 1957 recognises “computer generated works” under Section 2(d)(vi) ₁ and the “person who causes the work to be created” as the author. However, this provision was written well before the time of generative models. As a consequence, its applicability to modern AI systems (many of which are autonomous and unpredictable and do not have deterministic human control) has remained doctrinally unsettled.

Meanwhile, technological developments have exceeded the clarity of regulation. Corporations are utilizing AI in generating brand logos, marketing content, product design, architecture simulation and songs.Independent creators are now increasingly using AI to boost productivity or to produce new and original works. Yet, it is unclear who owns such outputs: Can they be copyrighted? If yes, under what conditions? If not, who is liable for infringement? This Article explores these tensions by critically assessing current legal structures, interpretations contained in the various doctrines, international judicial patterns and new policy initiatives. It forces India to adopt a balanced approach of innovation, which should retain the principles of human authorship and yet accommodate machine-assisted creativity. Generative AI refers to machine learning models – specifically large language models (LLMs) and generative adversarial networks (GANs) – that are able to generate original-looking content by detecting patterns in data sets that it is trained on. Unlike traditional software, which is guided by instructions written by humans, generative models make probabilistic decisions and create content that cannot even be predicted in advance by the developers.This autonomy presents a legal conundrum: the more sophisticated and creative the AI is the more difficult it is to claim authorship on a particular human. Courts have traditionally made it necessary for authorship to arise from a human intellect rather than an automated process[i].   

---

[i] Copyright Act, No. 14 of 1957, sec2 (d) (vi), India7

The Human Authorship Requirement

Copyright systems throughout the world insist that the works to be protected must be the result of human intellectual effort[i]. In India, the courts have repeatedly stressed on creativity, judgement and exercise of skill, which are inherently linked with human cognition. Thus, when AI generates creative outputs on its own, they may not be under the traditional boundary of copyright, and this generates important ownership questions:

Are the outputs of artificial intelligence copyrightable? If yes, who owns them? If not, do they belong to the public domain?

Why Generative AI Challenges the Doctrine of Copyright

Generative AI challenges the basics of the principles because:

• The output is not under the full control of humans.
• The system is often content generated from huge untraceable training data sets.  There is no legally-defined level of what constitutes “human involvement.”  AI could unintentionally copy and paste copyrighted material from its training data which could result in liability issues.
• Because the Indian Copyright Act is a pre-technology law, courts are left with interpretation until lawmakers bring the law into the modern era.

Legal Position of AI created work in India

Statutory Framework: Section 2(d)(vi) of the Indian Copyright Act refers to the author of a “computer generated work” as “the person who causes the work to be created.”[ii]This way of word indicates intentional human involvement. However, generative AI makes this requirement more difficult: who “causes” an AI-generated output? Possibilities include:

• the developer of the AI model,
• the person using the computer when he or she enters the prompt,  the operating company of the platform,  or no one at all (public domain).

This is not yet defined by Indian law.

Judicial Ambiguities: India does not have any case law that directly deals with generative AI ownership. Courts have handled software and the principles of creativity – but not autonomous machine production. Some of the relevant Indian precedents are:

• Modak (a little bit of creativity needed)[iii]
• R.G. Anand v. Deluxe Films (expression- not ideas)[iv]
• Eastern Book Co. v. Navin J. Desai (human mental efforts are essential)[v].Together, these suggest that works created by AI with no human creativity may not be protected by copyright and are therefore at risk of being copied.

Dangers of Unsecuring AI Outputs: If works created by AI are not copyrightable:

• Businesses may not be inclined to invest in AI-generated content.
• Competitors are free to copy the commercial outputs of AI outputs.
• Training datasets might be required to be disclosed in order to prevent

plagiarism claims.

• The economic value of AI-generated cultural and creative works may be lost

---

[i] U.S. Copyright Office, Compendium of copyright practices sec 306 (2021)8

[ii] Copyright Act, No. 14 of 1957, sec 2(d)(vi), India.

[iii] E. Book Co. v. D.B. Modak, (2008) 1 SCC 1

[iv] R.G. Anand v. Deluxe Films, (1978) 4 SCC 118.

[v] E. Book Co. v. Navin J. Desai, 2001 PTC 605 (Del.).9

Comparative International Approaching AI and Copyright

United States:One of the most explicit opinions on AI authorship is held by the

United States – AI-generated works may not be copyrighted unless they have a human contribution. This tenet was reiterated notably on numerous occasions by the U.S. Copyright Office (USCO)[ U.S. Copyright Office, Compendium of Copyright

Practices SS 306 (2021).] in guidance documents and rejection of registrations.. In the

Zarya of the Dawn graphic novel case[i], the USCO ruled that images created by

Midjourney could not be protected under copyright law because it was the AI system (as opposed to the human applicant) that determined the expressive elements of the image. The agency explained that human selection or arrangement of the outputs of artificial intelligence may be protectable, but the components of the output of the AI are not copyrightable. This approach strengthens a rigid requirement of human authorship. While this is a good way to protect traditional doctrine, it does leave industries that heavily use AI in the dark. Under this model, it is important for businesses using generative AI to include significant human creativity in order to get protections – something that Indian policymakers may need to think about.

United Kingdom: A hybrid model is adopted in the United Kingdom. Under

Section 9(3) of the UK Copyright, Designs and Patents Act (CDP) ₈ the author of a computer-generated work is considered to be “the person by whom the arrangements necessary for the creation of the work are undertaken.” This is similar to that of India’s Section 2(d)(vi) but is often interpreted broadly enough to include human involvement in processes of AI. UK courts have yet to define the precise standard of “arrangements” but commentators have suggested that prompt design, model training or system configuration may be enough. The UK model o5qffers greater certainty than the US approach and may shape the future reform in India as it explicitly considers non-human generation of works but still anchors authorship on the intention of human authors.

European Union: The EU is currently pursuing a harmonized regulatory plan through AI Act which classifies AI systems by the risk category and imposes obligations on developers and deployers. While the Act does not establish the rules of copyright, it does affect the way that the courts might consider AI autonomy, transparency and accountability. The European Parliament, in a number of occasions, has stated that copyright should also stay exclusive to humans, and rejected proposals for machine authorship. However, EU bodies are in favor of strong rights for human creators using AI, particularly with respect to derivative works. Additionally, the Court of Justice of the European Union (CJEU) never fails to emphasize “the author’s[ii]

own intellectual creation” as a standard for copyright.₉ It reaffirms human centered authorship and leaves open questions about complex, mixed-origin AI content.

What This Means for India: A review approach in the world sees three major trends:

• AI cannot be an author (U.S., EU).
• Human involvement is compulsory but undefined (UK, India).
• Hybrid or shared authorship models are emerging (academic proposals all over the world)
• India can use these models to create a flexible yet human-centred copyright regime.[iii]

Ownership Dilemmas in AI Created Works:

Who Is the Author? Possible Claimants:-

Because generative AI works at multiple levels of human and machine interaction, there are different parties that may potentially claim authorship:

• Developers: They design and train the AI model, and their contribution is great to its capabilities. However developers do not control individual outputs, and giving them ownership may lead to monopolies in all AI-generated works.₁₀
• Users (Prompt Creators): The user gives the input which triggers the output. But it can be found that courts consider prompts to be too minimal or generic to be considered creative contributions.
• Platform Operators: They host the model and may impact on its architecture. Yet their involvement can be too indirect to qualify for authorship.
• No One (Public Domain): This is the position of the United States: works that have been created purely by AI are granted no copyright, i.e. can be used by anyone. India needs to pick and choose or combine approaches that best suit itself and its creative economy and legal philosophy.₁₁

Human Oversight Model

Many scholars argue for a “human creative control” model – copyright should only attach when a human imposes a meaningful amount of control over the output. This may include:

• crafting detailed prompts,
• selecting or editing the outputs, and
• curating model behaviour,
• incorporating AI outputs into human created works.

Indian law could take such an approach through a statutory “substantial human contribution” requirement.

Joint Authorship Issues

Generative AI brings up the issue of shared authorship, especially when AI is being used to assist human creators. However the elements for joint authorship are:

• shared intent,[iv]
• significant autonomous contributions,  inseparable or interdependent employment.[v]

AI does not develop intent, and does not make legally recognized contributions. So, joint authorship between man and AI is doctrinally impossible under current law.

Copyrightability of Outputs of AI The major factors to be considered for India are:

• Fixation :AI outputs are digitally fixed and this requirement is met.
• Originality :Is AI output “creative”?
• Authorship : Is the human contribution enough?
• Liability : Who is liable for the infringements of AI?

Without the statutory revision, the Indian courts may find it difficult to answer these questions in coherent manner. (1213)

Legal & Ethical Challenges Created by Generative AI: Risks of Copyright Infringement:-

AI systems are trained on massive datasets that may have copyrighted works that have been scraped without permission. This raises questions of:

• unauthorized reproduction,
• derivative works,
• database rights,
• moral rights,
• and licensing obligations.[vi]

Even if the output of AI is not directly copied from a particular work, they may be substantially similar, which will leave users open to infringement claims.

Issues of Transparency and Explainability
Generative AI models are often “black boxes” – the inner workings of the model cannot be fully explained[vii]and it can be difficult to tell: whether the output contains a copyrighted material, how the training data was collected, what degree of human intervention played a part in creation. Indian courts can have a difficult time apportioning liability in the absence of explicit disclosure standards.

Moral Rights Concerns
Under Sections 57 of the Indian Copyright Act, authors are given the moral rights such as: the right of paternity, the right to integrity. These rights cannot logically apply to AI, which has no consciousness or reputation, but AI systems may cause violations of the moral rights of human authors by creating distorted versions of their work, which creates dilemmas in enforcement.

Economic and Ethical Problems
AI output may flood the markets with cheap content undermining: human authors, illustrators, musicians, and writers. The ethical issues are: the disappearance of human creativity, unemployment connected with automation, the homogenization of culture, biases incorporated into the product of AI.Balancing innovation and fairness is the key to the creative economy in India.

Reforming the Indian Copyright Law for the AI World

The Need for Legally Clear Statutes
The Indian Copyright Act was developed in the year 1957 when machine learning, neural networks, and generative AI did not exist. Although Section 2(d)(vi) refers to “computer-generated works,” this term no longer represents the complexity of modern AI systems, which can work without the predictable guidance of humans and can create works without human direction. The lack of the clarity leads to: uncertainty for businesses that use AI, judicial inconsistency, unenforceable claims of ownership, And vulnerability to infringement dispute. As AI becomes part of sectors like film, education, software design and entertainment, legislative reform is no longer an option it is crucial.[viii]

Potential Models for India to Consider
India can consider three different legal models:

A. Human-Centered Model (U.S./EU Style)
Copyright is only applicable where substantial human creativity is demonstrated. AI generated work is under public domain. Advantages:

• allows for traditional copyright doctrine to be preserved
• avoids the legalization of machines  avoids monopolies in output of AI Disadvantages:
• discourages investment into AI-generated content  allows valuable works to be unprotected

B. UK Style “Arrangements Necessary” Model
Copyright is for the person who undertakes arrangements to create necessary. Advantages:

• flexible and not restricted to technology
• has statutory recognition
• is in consonance with the existing Section 2(d)(vi) of India Disadvantages:
• ambiguity as to what constitutes “arrangements”
• may lead to litigation for minimal human input

C. Sui Generis Rights of the AI-Generated Works
A new category of IP rights for the exclusive outputs of AI, shorter duration and narrower scope. Advantages:

• Provides protection which is tailored to AI
• Eliminates the redefinition of authorship in traditional copyright law Disadvantages:
• Politically and legally complex;
• May bring into conflict with international treaties

India could have a hybrid model that maintains human authorship and under regulatory supervision, it could give limited rights to AI generated content.

Policy Recommendations Suggestions to Legislation.

1. Definitely Artificial Intelligence Generated Works.
   The Act should also offer the definition of: AI-generated work, generative model, human-assisted Creation autonomous AI creation.This will assist in grounding judicial interpretation.
2. Present a Significant Contribution of Humans Test.
   India would need to formalise that copyright can only prevail when a human input of expression, choice, criterion or inventiveness has been met. This does not allow AI to have the legal personhood, and safeguard human creativity. Niva ElkinKoren, Rethinking Creativity in the Age of the AI, 39 Berkeley Tech. L.J. 45, 6365 (2022).]
3. Create a Mandatory AI Transparency.
   The platforms must reveal: the sources of data that are being used to train, whether any copyrighted content was incorporated, how and when outputs can be similar to training data; This will assist courts to cope with infringement claims.
4. Establishing a Digital-Rights Tribunal.
   Considering the technicalities of the areas of AI disputes, India ought to set up a special tribunal to handle: Conflicts over copyright through AI, and the allocation of liability, regulation of data-use, and cross-border infringement of AI.

Technology and Ethics Recommendations.

• Develop Blockchain Based Provenance Tracking A registry of blockchain may help with the provenance of AI generated work: authorship claims, timestamps,derivative relationships. This would assist in strengthening the evidentiary reliability.
• Courts and Regulators Judges promote AI literacy judges, lawyers, and other officials must know: the architecture of generative models, training data systems, the limitations of the Machine learning.This would allow the courts to have a consistent reasoning.
• Developing Ethical and Inclusive AI Datasets: AI systems can cause and perpetuate discrimination and stereotypes through the bias in the output produced. It is significant that different, ethically gathered datasets should be ensured as the constitutional values should be preserved. UNESCO, AI and Ethics Report (2021).

Suggestions to the Practicing Industry.

• Editors ought to add human driven editing levels.
• AI should be applied to businesses and clarification on ownership through contractual means.
• There are plagiarism checking software that should be installed on media platforms to prevent the misuse of copyrighted materials.

Conclusion :

The concept of generative AI is altering the face of creativity and transforming authorship. This does not make it easy to navigate copyright law due to its capability to analyze information, produce works of expression, and copy and imitate the style of other works. India, as a technologically ambitious country, is yet to have any clear statutory guidance on how to resolve the ownership and infringement dilemmas presented by the generative AI tools. The outdated provisions of the Copyright Act (written in the pre-digital revolution) just can not cope with issues of autonomy, creativity or responsibility in AI products. Meanwhile, an overall lack of protection of works created by AI may serve as a deterrent to innovation and commercial sustainability of works, including the motivation to invest in innovative technologies. This study proposes that India should engage in moderate reform. The law must recognize the human authorship as the key value and offer the clarification of the areas of protection of the AIassisted works and establish the institutional incentives to manage the problem of infringement and liability. Comparative experience in the US, UK and EU shows that they are not the best models but have something to show the new IP ecosystem of India. India needs a more modernised law that will safeguard the creators, treat all people equally, and apply AI in a responsible and constitutional and ethical manner. This kind of strategy will assist India in becoming successful in the new age of digital authorship and technological advancement.

---

[i] Zarya of the Dawn, U.S. Copyright Office Correspondence (2023).

[ii] Copyright, Designs and Patents Act 1988, c. 48, sec 9(3) (UK).10

[iii] European Parliament, Resolution on Intellectual Property Rights for AI (2020).11

[iv] Pamela Samuelson, Allocating Ownership of AI Outputs, 67 J. Copyright Soc’y 321, 329 (2020).

[v] U.S. Copyright Office, Policy Statement on AI (2023).

[vi] UK IPO, AI and IP: Copyright Issues (2022).

[vii] OECD, AI in Society 61-63 (2019).14

[viii] Copyright Act, No. 14 of 1957, sec 2(d)(vi), India.