Copyright And Generative Artificial Intelligence: Liability for Training Vs Output in India

Jashanpreet Singh Thind
Rayat Bahra University, Mohali

Ms. Priyanka Dhiman
Assistant professor
Rayat Bahra University, Mohali

Abstract
This Study examines the evolving relationship between copyright law and generative artificial intelligence in India, with a focused inquiry into liability at two legally distinct stages: (i) the training of AI models on copyrighted works, and (ii) the generation and dissemination of AI outputs that may reproduce or closely resemble protected expression. It analyses how the Copyright Act, 1957—particularly the framework on subsistence and scope of rights, infringement, and statutory exceptions—applies to modern AI workflows involving large-scale electronic copying, storage, and computational processing. The study argues that “training-stage” conduct raises foundational questions about whether dataset compilation and model training amount to reproduction and other restricted acts, and whether existing exceptions can legitimately accommodate industrial-scale use. It further examines “output-stage” risks, including reproduction/adaptation concerns and attribution and integrity interests, and evaluates how responsibility may be distributed among developers, deployers, platforms, and end users. The dissertation identifies doctrinal and regulatory gaps arising from the mismatch between traditional copyright assumptions and contemporary AI systems, and proposes principles for clearer statutory interpretation and targeted reforms, including structured licensing approaches, transparency obligations, and proportionate safeguards to protect creators while enabling innovation.

Keywords: Generative AI; Copyright Act, 1957; Training Data; AI Outputs; Infringement Liability; Statutory Exceptions

1 Introduction
Generative artificial intelligence has rapidly moved from experimental research to mass consumer and enterprise deployment, enabling systems to produce text, images, music, code, and audiovisual content that appears creative and human-like. This shift poses a direct challenge to copyright law because the core technical ingredient of modern generative models is large-scale learning from existing human expression, much of which is protected by copyright. In India, this tension is especially significant because the Copyright Act, 1957 was drafted for human authorship and traditional modes of reproduction, yet it must now be applied to computational training pipelines and algorithmic outputs distributed at internet scale. The central legal question is not whether AI is “good” or “bad,” but how Indian copyright doctrine allocates responsibility and risk between those who build models, those who deploy them, and those who use them—particularly at two stages where copyright conflicts arise: first, when models are trained on protected works, and second, when the system produces outputs that may replicate, adapt, or closely imitate those works. [1]

Indian copyright law begins with the idea that copyright subsists in specific classes of works and grants exclusive rights to the owner, which become the doctrinal anchor for assessing AI-related conduct. Section 13 of the Copyright Act, 1957 identifies the principal categories of protected subject matter, including original literary, dramatic, musical, and artistic works, as well as cinematograph films and sound recordings, all of which are routinely used as training data and are also the types of content that generative AI can output. Section 14 then defines “copyright” as a bundle of exclusive rights (such as reproduction and other exploitations) that are subject to the Act’s limitations and exceptions, meaning that any AI practice that functionally entails copying or storing protected expression will be tested against these exclusive rights and the permitted acts regime. For generative AI, the legal analysis cannot remain abstract: training is technically dependent on making copies (often temporary, cached, or stored) and extracting patterns from datasets, while the output stage is about what the system communicates to the user and how closely it tracks protected expression. The dissertation therefore treats “training liability” and “output liability” as analytically distinct but practically connected questions under the same statutory architecture. [2]

1.1 Background and Contemporary Relevance of Generative Artificial Intelligence
Generative artificial intelligence refers to computational systems that can produce new content—such as text, images, audio, video, or code—based on patterns learned from large datasets. Unlike earlier “rule-based” software, modern generative AI typically relies on machine-learning architectures that learn statistical relationships between symbols and features, enabling the system to generate outputs that can appear meaningfully “creative” to ordinary users. This practical capacity has made generative AI a general-purpose technology for education, entertainment, marketing, programming, design, and legal drafting, creating significant economic and cultural impact. Yet its ability to generate expressive material also places it in direct contact with copyright’s domain, because copyright primarily protects original expression in literary, dramatic, musical and artistic works, and extends to films and sound recordings—precisely the forms of content that generative AI systems are increasingly trained on and can output. [1] In India, the contemporary relevance of generative AI is amplified by rapid digitisation, platform-based distribution of creative works, and the integration of AI tools into everyday productivity and media consumption. Public policy institutions have described AI as a strategic lever for economic and social transformation, indicating that India is likely to witness deeper adoption of AI systems across sectors. This makes it essential to analyse whether existing legal frameworks—especially the Copyright Act, 1957—are capable of addressing AI-driven creation and dissemination without undermining either innovation or authors’ rights. The Copyright Act is the central statute governing copyright in India, and its core structure—copyright subsistence in protected classes of works (Section 13), the meaning and scope of exclusive rights (Section 14), and infringement rules (Section 51)—is the starting point for assessing the legality of AI-related acts that involve copying or generating expressive content.

1.3 Objectives of the study

1. To investigates authorship The study investigates authorship through copyright law which was established in 1957 Copyright Act 1957 with its Sections 2d and 13.
2. To study how AI-generated works create new challenges for existing legal systems which define human authorship according to their copyright ownership regulations.
3. To assess whether current Indian copyright laws effectively protect authorship and ownership rights for AI-generated content through sections 14 and 17 of the copyright system.
4. To investigates judicial and doctrinal challenges which emerge when people try to understand computer-generated authorship according to Section 2d(vi) which requires case-based evidence.
5. To proposes legal and policy reforms which will resolve authorship and ownership disputes for AI-generated works through Sections 18 and 19 statutory transfer rules and Section 52 public-interest balancing.

1.4 Research Questions

1. Whether and to what extent existing global legal frameworks recognise authorship in relation to AI-generated works?
2. To what extent do different jurisdictions differ in attributing authorship where no human creative control is involved?
3. Whether the concepts of authorship and originality under Sections 2(d) and 13 of the Copyright Act, 1957 can accommodate AI-generated works within the present Indian regime?
4. What doctrinal ambiguities and interpretive challenges arise in India when determining authorship of AI-generated works in the absence of explicit statutory recognition?
5. Whether legislative amendments or interpretive development are necessary to resolve uncertainties surrounding authorship and ownership of AI-generated works in India, especially under Sections 14, 17, 18 and 19?

1.5 Research Methodology
The research methodology adopted in this study is doctrinal because it systematically studies legal principles and statutory interpretation and secondary authoritative materials which explain copyright and AI-generated works. The study uses the Copyright Act 1957 as its main source and investigates essential sections that define authorship and ownership through Sections 2(d), 2(o), 2(y), 13, 14, 17, 18, 19, 52 and 57 to determine whether these sections permit AI-generated works under current legal standards. The doctrinal method requires researchers to read statutory language closely while they uncover hidden beliefs that the Act contains especially its human-focused approach to creativity and they perform legal analysis based on how the Act presents its authorship rights and transfer rights distribution system.

1.5 Review of Literature
Toby Bond & Sarah Blair, (2019)[1] article reviews the UK system for computer-generated content and demonstrates how section 9(3) attributes authorship to works which lack human creators. The study identifies two primary problems which include the difficulty of distinguishing between human-written and machine-produced content and the ongoing need to determine whether such content meets originality requirements. The document serves as a helpful resource for comparing different legal systems which determine authorship rights of works created by artificial intelligence.

Söğüt Atilla (2024)[2] revisits section 9(3) of the Copyright, Designs and Patents Act 1988 to demonstrate that modern AI-generated content requires the provision to be evaluated anew. The article provides its main value through its examination of current law which establishes how statutory language should be interpreted and how that interpretation should impact present copyright reform discussions.

Artha Dermawan (2024)[3] shifts the AI-copyright debate away from a purely owner-centred framework and argues that public interest theory should guide the balance between rightsholders and society in the generative AI era. The article is important for showing that ownership questions should not be resolved only through incentives, but also through access, dissemination of knowledge, and cultural participation.

Johannes Fritz (2024)[4] article studies the concept of authorship under EU law and explains that EU legislation does not clearly define who may be an author or what exactly qualifies someone as one. Fritz uses the doctrinal gap in his research to examine AI-related uncertainties while also establishing a solid theoretical base for future academic discussions about whether non-human and semi-machine-generated works should be considered copyright authorship.

Johannes Fritz (2025)[5] examines AI-assisted works rather than fully autonomous AI outputs and focuses on the blurred boundary between human and machine contribution. The article asks whether authorship still exists in such works and, if it does, to whom it should be attributed. Its structured approach is especially useful for analysing degrees of human control in AI-assisted creativity.

2 Technological and Legal Foundations
2.1 Technical Architecture of AI Systems: Data, Training, Prompts and Outputs
The technical lifecycle of a generative AI system typically begins with data acquisition and dataset construction, where large corpora of text, images, audio, or code are collected, cleaned, deduplicated, and converted into standardized formats suitable for training. This stage is legally sensitive in India because data acquisition often involves copying and storing materials, and Section 14 explicitly treats electronic storage as part of reproduction for multiple categories of works. Since many training corpora contain works protected under Section 13, the process of collecting and storing them can implicate exclusive rights under Section 14 and become actionable through Section 51 if done without licence, unless the dataset use fits within one of Section 52’s listed permitted acts. Therefore, even before model training begins, the architecture of dataset creation already maps onto copyright’s reproduction and storage concepts in Indian law. [6]

2.2 Functional Difference Between Generative AI and conventional software systems
Conventional software systems typically operate through deterministic or semi-deterministic logic: developers specify rules, and the program applies those rules to inputs to generate outputs that are intended and predictable within defined parameters. Generative AI systems, by contrast, operate through learned statistical representations and generate outputs probabilistically, meaning the same prompt can produce different outputs and the system’s behaviour is shaped by training data rather than only by programmer-written instructions. This functional difference matters for Indian copyright because the legal system must identify which human actors performed legally relevant acts—copying, storing, issuing copies, communicating to the public—under Section 14, and whether those acts are attributable to the developer, deployer, or user for Section 51 purposes. Determinism is not required for infringement; rather, the question is whether a restricted act occurred without licence, and the architecture of generative AI changes where and how those acts occur (especially at the training stage, where copying and storage are necessary for learning). [7]

2.3 Fundamental Principles and Objectives of Copyright Law
Copyright law is a statutory framework designed to protect original expression and allocate exclusive rights to creators (or rights-holders) to encourage creation, dissemination, and investment in cultural and informational works. In India, these principles are reflected through Section 13, which establishes the works in which copyright subsists, and Section 14, which defines copyright as a bundle of exclusive rights subject to the Act’s provisions. The core objective is to provide a legally enforceable mechanism that prevents unauthorized reproduction and exploitation while still allowing socially valuable uses through limitations and exceptions. This balance is operationalized through Section 51 (which defines when infringement is deemed to occur) and Section 52 (which specifies acts that do not constitute infringement), making the Indian statute a structured balancing mechanism rather than an absolute monopoly. [8]

3 Indian Copyright Framework (Copyright Act, 1957)
3.1 Scheme and Structure of the Copyright Act, 1957
The Copyright Act of 1957 operates as an author-based law which establishes copyright protection for specific types of works and distributes rights to those works while the law governs ownership and transfer and enforcement through a connected sequence of legal statutes. The Act begins with definitional provisions that classify what counts as a “work” and who counts as an “author,” then provides rules on when copyright subsists, what rights it includes, who owns it in the first instance, and how it may be transferred or limited. The AI-generated work system requires this structure because all subsequent matters which include licensing and infringement and remedies and registration processes depend on the law determining which works receive protection and which authors possess legal recognition. The Act defines Section 2(y) as “work” which includes literary and dramatic and musical and artistic works and cinematographic films and sound recordings. The Act defines “literary work” in Section 2(o) to include computer programs together with tables and compilations that contain computer databases.[9]

3.2 Meaning of Author under the Indian Copyright Framework
The Copyright Act 1957 defines “author” in Indian copyright law through Section 2(d) which offers specific definitions for each work instead of providing one common definition. The author of a literary or dramatic work is the person who creates it; the composer is the author of a musical work; the artist is the author of an artistic work; and in the case of certain subject matter such as films and sound recordings, authorship is linked to the producer. The method demonstrates how the statute establishes authorship as a legally recognized function which connects to both creative processes and the right to control how a work gets presented. [10]

Section 2(d)(vi) provides the most important section for works created by artificial intelligence because it defines computer-generated works as literary dramatic musical or artistic works which belong to the person who creates them. This serves as the legal foundation for AI discussions in India because it describes computer-based creation while asserting that authorship must belong to human beings instead of machines. The term “person who causes” includes multiple actors which results in confusion when applying it to current generative artificial intelligence systems.

4 Judicial Landscape and Comparative Legal Approaches
4.1 Need for Comparative Analysis in AI Copyright Governance
The authorship of AI-generated works has become an international matter which requires comparative analysis to establish effective AI copyright regulations. The development and training and operation and licensing and usage of generative AI systems occur across international boundaries because their generated content becomes available through worldwide digital distribution systems. Copyright systems which handle AI creation issues need to establish ownership rights and licensing procedures and enforcement methods because works created in one region and used in another and stored on international platforms create confusion about ownership rights.[11] The Copyright Act of 1957 which still applies in India through its Sections 2(d) and 13 and 14 and 17 active human-centred model requires investigation because other countries search for methods to regulate AI-created content. Comparative analysis therefore helps identify whether India’s approach under Section 2(d)(vi), which attributes authorship in computer-generated works to “the person who causes the work to be created,” is conceptually aligned with wider legal developments or whether it is becoming insufficient for contemporary generative AI. [12]

4.2 International Copyright Principles Relevant to Authorship
International copyright laws do not designate AI technology for copyright authorship rights yet their established rules remain crucial to determining who holds copyright. The Berne Convention for the Protection of Literary and Artistic Works provide author rights protection through its framework which secures the rights of authors to their protected works. The document requires existing member states to protect works through national treatment and minimum standards which enable literary and artistic works to remain protected by recognized authorship. Although the treaty does not provide a specific definition of author that applies to AI-created content the treaty designates authorship rights to human creators instead of machines. The Copyright Act 1957 requires Indian copyright law to be interpreted according to treaty obligations when determining whether Sections 2(d) and 13 can extend to cover AI-generated content. [13]

Copyright law protects creative expression according to the second fundamental international copyright rule which maintains that no protection exists for ideas or operational methods or mathematics. The TRIPS Agreement Article 9(2) establishes this rule which directly impacts AI authorship because AI systems depend on algorithms and model weights and operational procedures and computational methods which do not qualify for copyright protection of their created expressive content. The copyright system protects creative works that emerge from human-based or legally recognized creative activities through their expressed content. The principle restricts copyright claims which depend solely on the output of an artificial intelligence system to produce original work. The authorship analysis requires researchers to trace back all expressions to their actual sources of creation instead of studying all technological methods that generate recorded works. The principle of this law interacts with Section 13 of the Copyright Act 1957 which establishes originality requirements for literary works and dramatic works and musical works and artistic works according to Indian law.[14]

4.3 Position of Jurisdictions Recognising Human-Centred Authorship
A substantial number of legal jurisdictions still maintain their recognition of a human-cantered authorship model, which establishes copyright rights through human artistic work instead of acknowledging AI systems as creators. The United States stands as one of the most apparent modern-day examples. The U.S. Copyright Office’s 2023 registration guidance and its 2025 Part 2 Report on Copyrightability state that copyright does not extend to material generated wholly by AI without sufficient human authorship, although human-authored selection, arrangement, or modifications may still be protected if they meet the ordinary threshold of originality. The established position maintains the traditional connection between copyright protection and human creative work while denying that prompting activities establish ownership rights over resultant outcomes. The U.S. approach is therefore strongly human-centred and provides a useful comparative contrast to India’s more open-ended wording in Section 2(d)(vi). [15]

Numerous European legal systems demonstrate their support for human-cantered authorship. German copyright law begins from the principle that authors of works in the literary, scientific and artistic domain enjoy protection for their works, reflecting the continental author’s-right tradition. French law states in Article L111-1 of the Code de la propriété intellectually that the author of a work of the mind enjoys, by the sole fact of creation, an exclusive incorporeal property right in that work. Spanish law similarly provides in Article 5 that the individual who creates any literary, artistic or scientific work shall be deemed the author thereof. The author emerges as the foundational human creator through these formulations which support civil-law theories that copyright exists because people create original works through their intellectual efforts instead of machines producing copyrighted materials. Indian law needs to consider this comparative trend because multiple systems which uphold strong moral-rights traditions maintain their requirements that human authorship must remain intact in today’s world of artificial intelligence. [16]

4.4 Case Laws
Eastern Book Company & Ors. v. D.B. Modak & Anr.[17] The Supreme Court decision serves as the primary basis for evaluating copyright originality under Indian law Copyright protection in India requires more than basic effort because it needs originality to demonstrate substantial intellectual work.

R.G. Anand v. Delux Films & Ors.[18] The case establishes copyright existence to safeguard artistic expression but not the underlying conceptswhich fundamental to evaluating AI-generated textual content and visual materials and audiovisual creation that may exhibit identical themes or artistic styles.

Syndicate of the Press of the University of Cambridge v. B.D. Bhandari & Anr[19] The decision functions as a reference point which scholars use to examine copyright enforcement and defense mechanisms while it provides insights into how courts evaluate copyright infringement cases through analysis of ownership disputes and copyright exemptions established in the 1957 Copyright Act. Indi

Civic Chandran & Ors. v. C. Ammini Amma & Ors.[20] The case holds importance because it establishes that copyright protection needs to work together with public critique and public interest which increases the effectiveness of Section 52 fair dealing provisions in India. Modern generative systems used for AI-generated works create criticism and parody and commentary which makes this issue important for the industry.

Thaler v. Perlmutter[21] This U.S. decision represents the current top judicial ruling about AI-generated works which AI systems produced without any human artistic contribution. The decision helps India because it demonstrates that machine-generated content receives copyright protection only when it has human authorship.

Naruto v. Slater[22] The “non-human claimant” decision has a major impact on current AI discussions because it shows that U.S. copyright law recognizes authorship only for legal persons who own copyrights and not for non-human entities.

Feist Publications, Inc. v. Rural Telephone Service Co[23] The case of Feist establishes the basic legal standard which requires that original work must be created completely from new material which contains at least a small degree of creative expression.

The case of Burrow-Giles Lithographic Co v. Saron[24] establishes that copyright law protects original intellectual concepts which photographers express through their work. The authorship of works depends on human beings who select specific artistic elements according to copyright law. The Indian legal system needs to interprete Section 2dvi through creative control standards whereas Section 13 originality should serve as a test to identify non-mechanical content.

4.5 Legal Challenges

4.5.1 Nature of Legal Uncertainty in Relation to AI-Generated Works
The Copyright Act of 1957 establishes legal uncertainty for AI-generated works in India because its authors based creative work on human artistic ability, which does not apply to modern AI tools that create content through statistical predictions and machine-driven methods. The statutory chain—classification of “work” (Section 2(y)), subsistence and originality (Section 13), authorship (Section 2(d)), ownership (Section 17) and exclusive rights (Section 14)—depends on the law’s ability to identify a legally relevant author. [25]

4.5.2 Challenges in Identifying the True Author of AI-Generated Content
The process of discovering who actually created the AI-generated material becomes difficult because generative AI systems share creative responsibility through various phases of their operation. The process involves four distinct roles: a developer builds the model, a deployer manages its setup and performance, a user gives input and chooses results, and an editor improves the completed product. Indian copyright law requires that a legally recognizable author needs to exist under Section 2(d) before any copyright owner can obtain first rights under Section 17 and all associated rights under Section 14. The legal system lacks a defined structure which identifies who among multiple creators of artificial intelligence works deserves the title of author when machines primarily determine the finished product. [26]

4.5.3 Absence of Explicit Statutory Recognition of AI Authorship
Indian law does not recognize AI systems as authors according to its existing legal framework. The definition of “author” in Section 2(d) limits its application to human or human-connected roles which include author, composer, artist, photographer, and producer while the computer-generated clause in Section 2(d)(vi) designates authorship to a “person” instead of a machine. The current law states that AI systems lack legal capacity to create written works because all AI-generated content needs to be assigned to either human creators or legal entities. Courts face uncertainty about AI authorship because legal frameworks require them to decide between two options: (i) establishing authorship through human representation as stated in Section 2(d)(vi) or (ii) refusing protection when human creative control does not exist. [27]

5 Conclusion and Recommendations
5.1 Conclusion
International copyright systems do not recognize artificial intelligence systems as authors because these systems validate human creators who possess legal status as the only valid authors. The Berne Convention establishes author protection for literary and artistic works through its main structure which estimates authorship rights to known authors, even though the convention lacks a current AI definition of authorship. The jurisdictions show significant differences because they have two main systems which exist between (i) strict human-authorship systems that deny copyrightability to fully autonomous outputs and (ii) proxy-human attribution systems that preserve protection by attributing authorship to a person connected to the generation process. The United Kingdom’s Copyright Designs and Patents Act 1988 serves as the most straightforward statutory proxy model because section 9(3) establishes authorship for computer-generated literary dramatic musical or artistic works as belonging to the person who organizes all elements required for their creation.

The United States system establishes a human-authorship threshold which determines that all AI-generated content lacks copyright protection except for human-written elements which include original selection and arrangement and modification. Japan’s official guidance treats AI-generated materials as non-protectable works while treating AI itself as an entity that cannot be considered an author but it does accept protection for situations where AI functions as a creative tool under human direction and contribution. The European Union establishes its originality requirements through directives which define copyrightable works as the “author’s own intellectual creation” thus making it impossible to recognize autonomous outputs as protected works. The majority of jurisdictions reach the same conclusion because they assess works without human input for copyright protection except for the United Kingdom which maintains its distinct approach to computer-generated works through its statutory framework which creates a different legal environment for those creations.

The Indian statute provides partial yet incomplete elements to accommodate modern generative AI technologies which lack complete integration. Section 2(d)(vi) of the law specifically includes computer-generated literary dramatic musical and artistic works which create authorship rights for the person who generates the work because the Indian government created this law framework to recognize computer-based creation and protect authorship rights through human or legal person identification as authorship holders.

The Act creates its statutory partial accommodation through its operational sequence which starts with work classification in Section 2(y) Section 2(o) and continues through subsistence and originality in Section 13 authorship in Section 2(d) exclusive rights in Section 14 first ownership in Section 17 and transfers in Sections 18–19. AI-generated works challenge the chain at multiple points simultaneously.

The primary ambiguity in Indian law exists because Section 2(d)(vi) lacks statutory text while Section 2(d)(vi) creates uncertainty about its meaning and statutory criteria for determining originality remain undefined under Section 13. The phrase “person who causes the work to be created” has two interpretations which include a narrow meaning that identifies the creative controller of the work and a wide definition which recognizes the person who begins the production process.

5.2 Recommendations

1. The legislative amendment together with the official interpretive guidance will determine the definition of computer-generated content and the necessary creative control level which authors must achieve for their work to qualify as Section 2(d)(vi) authorship.
2. The law should explicitly distinguish AI-assisted works (where a human meaningfully shapes expression) from autonomous AI outputs (where the system determines expression with minimal human control).
3. The statutory formulation requires that authorship attribution under Section 2(d)(vi) needs the creator of the work to show major artistic control through methods which include structured prompting and iterative refinement and selection of elements which demonstrate their intellectual ability and extensive editing which creates expressive components.
4. The ownership effects of Section 17 require explanation through AI-generated works which multiple parties create. The law maintains its standard that authorship gives initial ownership to authors but needs to establish AI-specific presumptions which include two scenarios.
5. The goal of this document is to establish administrative guidelines which will bring registration practice into alignment with evidentiary standards for authorship and ownership declarations in application submissions.
6. The public interest needs to maintain its equilibrium because any expansion about AI-based work protection should follow Section 52 exceptions and prevent additional restrictions on machine-created content. The protection of AI outputs which require minimal human contribution would result in broadening Section 14 exclusive rights across educational research and critical evaluation together with other activities that benefit society.

Bibliography
Statutes

1. The Copyright Act, 1957
2. The Copyright Rules, 2013
3. The Information Technology Act, 2000
4. The Digital Personal Data Protection Act, 2023
5. Berne Convention for the Protection of Literary and Artistic Works, Sept. 9, 1886, as revised,
6. Agreement on Trade-Related Aspects of Intellectual Property Rights, Apr. 15, 1994

Books

1. Ryan Abbott, THE REASONABLE ROBOT: ARTIFICIAL INTELLIGENCE AND THE LAW 1 (2020), https://thereasonablerobot.com/about-the-author/.
2. Ryan Abbott ed., RESEARCH HANDBOOK ON INTELLECTUAL PROPERTY AND ARTIFICIAL INTELLIGENCE 1 (2022), https://www.elgaronline.com/edcollbook/book/9781800881907/9781800881907.xml.
3. Aviv H. Gaon, THE FUTURE OF COPYRIGHT IN THE AGE OF ARTIFICIAL INTELLIGENCE 1 (2021), https://www.elgaronline.com/monobook/9781839103148.xml.
4. Eleonora Rosati, COPYRIGHT IN THE DIGITAL SINGLE MARKET: ARTICLE-BY-ARTICLE COMMENTARY TO THE PROVISIONS OF DIRECTIVE 2019/790 63 (2021), https://global.oup.com/academic/product/copyright-in-the-digital-single-market-9780198858591.
5. Paul Goldstein & P. Bernt Hugenholtz, INTERNATIONAL COPYRIGHT: PRINCIPLES, LAW, AND PRACTICE 207 (4th ed. 2019), https://academic.oup.com/oxford-law-pro/book/60050.

Journal Articles

1. Toby Bond & Sarah Blair, Artificial Intelligence & copyright: Section 9(3) or authorship without an author, 14 J. INTELL. PROP. L. & PRAC. 423, 423 (2019), https://doi.org/10.1093/jiplp/jpz056.
2. Söğüt Atilla, Dealing with AI-generated works: lessons from the CDPA section 9(3), 19 J. INTELL. PROP. L. & PRAC. 43, 43 (2024), https://doi.org/10.1093/jiplp/jpad102.
3. Artha Dermawan, AI v copyright: how could public interest theory shift the debate on ownership in the age of generative AI?, 19 J. INTELL. PROP. L. & PRAC. 55, 55 (2024), https://doi.org/10.1093/jiplp/jpad111.
4. Johannes Fritz, The notion of ‘authorship’ under EU law—who can be an author and what makes one an author? An analysis of the legislative framework and case law, 19 J. INTELL. PROP. L. & PRAC. 552, 552 (2024), https://doi.org/10.1093/jiplp/jpae022.
5. Johannes Fritz, Understanding authorship in Artificial Intelligence-assisted works, 20 J. INTELL. PROP. L. & PRAC. 354, 354 (2025), https://doi.org/10.1093/jiplp/jpae119.
6. Daryl Lim, Generative AI and copyright: principles, priorities and practicalities, 18 J. INTELL. PROP. L. & PRAC. 841, 841 (2023), https://doi.org/10.1093/jiplp/jpad081.
7. Kalpana Tyagi, Copyright, text & data mining and the innovation dimension of generative AI: a three-step test assessment, 19 J. INTELL. PROP. L. & PRAC. 557, 557 (2024), https://doi.org/10.1093/jiplp/jpae028.
8. Nicola Lucchi, ChatGPT: A Case Study on Copyright Challenges for Generative Artificial Intelligence Systems, 15 EUR. J. RISK REGUL. 602, 602 (2024), https://doi.org/10.1017/err.2023.59.
9. Eleonora Rosati, Infringing AI: Liability for AI-Generated Outputs under International, EU, and UK Copyright Law, 16 EUR. J. RISK REGUL. 603, 603 (2025), https://doi.org/10.1017/err.2024.72.
10. Mirko Degli Esposti, Francesca Lagioia & Giovanni Sartor, The Use of Copyrighted Works by AI Systems: Art Works in the Data Mill, 11 Eur. J. RISK REGUL. 51, 51 (2020), https://doi.org/10.1017/err.2019.56.

---

[1] Toby Bond & Sarah Blair, Artificial Intelligence & copyright: Section 9(3) or authorship without an author, 14 J. INTELL. PROP. L. & PRAC. 423, 423 (2019), https://doi.org/10.1093/jiplp/jpz056

[2] Söğüt Atilla, Dealing with AI-generated works: lessons from the CDPA section 9(3), 19 J. INTELL. PROP. L. & PRAC. 43, 43 (2024), https://doi.org/10.1093/jiplp/jpad102

[3] Artha Dermawan, AI v copyright: how could public interest theory shift the debate on ownership in the age of generative AI?, 19 J. INTELL. PROP. L. & PRAC. 55, 55 (2024), https://doi.org/10.1093/jiplp/jpad111

[4] Johannes Fritz, The notion of ‘authorship’ under EU law—who can be an author and what makes one an author? An analysis of the legislative framework and case law, 19 J. INTELL. PROP. L. & PRAC. 552, 552 (2024), https://doi.org/10.1093/jiplp/jpae022

[5] Johannes Fritz, Understanding authorship in Artificial Intelligence-assisted works, 20 J. INTELL. PROP. L. & PRAC. 354, 354 (2025), https://doi.org/10.1093/jiplp/jpae119

[6] Ali Arsanjani, “The Generative AI Life-cycle” Medium, 2023 available at: https://dr-arsanjani.medium.com/the-generative-ai-life-cycle-fb2271a70349 (last visited May 13, 2026).

[7] Akash Takyar, “Generative AI – A beginner’s guide” Leewayhertz, 24 April 2023.

[8] “Copyright Law in India: Meaning, History, Objectives, Nature & Copyright Infringement,” available at: https://thelegalschool.in/blog/copyright-law-in-india (last visited May 13, 2026).

[9] Copyright Act, 1957 IPLEADERS, https://blog.ipleaders.in/an-overview-of-the-copyright-act-1957/ (last visited Apr 23, 2026)

[10] Ownership of AI generated content. A Deep Dive into Copyright Law in India KHURANA AND KHURANA, https://www.khuranaandkhurana.com/ownership-of-ai-generated-content-a-deep-dive-into-copyright-law-in-india (last visited Apr 23, 2026)

[11] Hafiz Gaffar & Saleh Albarashdi, Copyright Protection for AI-Generated Works: Exploring Originality and Ownership in a Digital Landscape, 15 ASIAN JOURNAL OF INTERNATIONAL LAW 23–46

[12] Ownership and Authorship of AI- Generated Works under Indian Copyright law: A Regulatory Gap Analysis VINTAGE LEGAL, https://www.vintagelegalvl.com/post/ownership-and-authorship-of-ai-generated-works-under-indian-copyright-law-a-regulatory-gap-analysi (last visited Apr 5, 2026)

[13] BERNE CONVENTION FOR THE PROTECTION OF LITERARY AND ARTISTIC WORKS TREATIES, https://www.wipo.int/en/web/treaties/ip/berne/index (last visited Apr 5, 2026)

[14] WTO INTELLECTUAL PROPERTY – OVERVIEW OF TRIPS AGREEMENT, https://www.wto.org/english/tratop_e/trips_e/intel2_e.htm (last visited Apr 5, 2026)

[15] Hafiz Gaffar & Saleh Albarashdi, Copyright Protection for AI-Generated Works: Exploring Originality and Ownership in a Digital Landscape, 15 ASIAN JOURNAL OF INTERNATIONAL LAW 23–46

[16] ACT ON COPYRIGHT AND RELATED RIGHTS (URHEBERRECHTSGESETZ – URHG), https://www.gesetze-im-internet.de/englisch_urhg/englisch_urhg.html (last visited Apr 5, 2026)

[17] Eastern Book Company & Ors. v. D.B. Modak & Anr., (2008) 1 SCC 1; AIR 2008 SC 809

[18] R.G. Anand v. Delux Films & Ors., (1978) 4 SCC 118; AIR 1978 SC 1613

[19] Syndicate of the Press of the University of Cambridge v. B.D. Bhandari & Anr., Delhi High Court (DB), 185 (2011) DLT 346

[20] Civic Chandran & Ors. v. C. Ammini Amma & Ors., 1996 (1) KLT 608; 16 PTC 329 (Ker.)

[21] Thaler v. Perlmutter, 687 F. Supp. 3d 140 (D.D.C. 2023); aff’d, U.S. Court of Appeals (D.C. Cir.) No. 23-5233

[22] Naruto v. Slater, 888 F.3d 418 (9th Cir. 2018)

[23] Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340 (1991)

[24] Burrow-Giles Lithographic Co v. Sarony 111 U.S. 53 1884

[25] AI-Generated Work’s Protection Under The Copyright Act, 1957 IJLLR JOURNAL, https://www.ijllr.com/post/ai-generated-work-s-protection-under-the-copyright-act-1957 (last visited Apr 8, 2026)

[26] Akash Takyar, How to build a generative AI solution?, LEEWAYHERTZ, Feb. 28, 2023, https://www.leewayhertz.com/how-to-build-a-generative-ai-solution/ (last visited Apr 8, 2026)

[27] Hafiz Gaffar & Saleh Albarashdi, Copyright Protection for AI-Generated Works: Exploring Originality and Ownership in a Digital Landscape, 15 Asian JOURNAL OF INTERNATIONAL LAW 23–46

---

[1] “Generative AI: Revolutionizing technology,” ELO Digital Office GmbH available at: https://www.elo.com/en-de/blog/generative-ai.html (last visited May 13, 2026).

---

[1] “A Systematic Literature Review on the Impact of Generative AI in Digital Marketing: Advancements, Opportunities, and Challenges,” 87 Computers, Materials and Continua.

[2] seo835, “Ownership of AI generated content. A Deep Dive into Copyright Law in India” Khurana And Khurana, 2025 available at: https://www.khuranaandkhurana.com/ownership-of-ai-generated-content-a-deep-dive-into-copyright-law-in-india (last visited May 13, 2026).