Challenges Of Copyright Protection For AI- Generated Content - A Comparative Study Of India And Australia

Author: LawSphereX

Category: Intellectual Property

Published: 2026-08-23T18:46:07.009Z

Updated: 2026-08-30T07:07:58.617Z

CHALLENGES

OF COPYRIGHT PROTECTION FOR AI- GENERATED CONTENT: A COMPARATIVE STUDY OF INDIA AND AUSTRALIA

Anshika Kapoor, B.B.A. LL.B., Bharati Vidyapeeth University

ABSTRACT

The emergence of generative artificial intelligence systems capable of producing literary, artistic, musical, and audiovisual works of considerable sophistication has exposed a structural fault line in the architecture of copyright law as it exists in virtually every jurisdiction in the world. Copyright doctrine is built on the premise of human authorship: the author is the natural person whose intellectual creativity gives rise to the work, and the rights the law grants are conceived as a form of recognition and incentive directed at that creative person. Generative AI disaggregates this premise by introducing a technological intermediary whose outputs are produced through processes that may involve no human creative decision-making at the point of generation, even where the system was trained on humanauthored works and is operated through human-crafted prompts.

This paper undertakes a comparative analysis of how the copyright frameworks of India and Australia two common law jurisdictions with shared colonial legislative heritage but diverging approaches to the challenge of computer-generated works address this problem, and what each jurisdiction's approach reveals about the adequacy of existing copyright doctrine in the age of generative AI. It examines the statutory definitions of authorship in the Copyright Act,

1957 (India) and the Copyright Act, 1968 (Cth) (Australia), the judicial

development of originality standards in both jurisdictions, the treatment of computer-generated works under each Act, and the emerging questions regarding ownership, moral rights, and the use of copyright works for AI training. The paper identifies three central challenges: the authorship gap, which arises when AI generates works without sufficient human creative contribution to satisfy the originality requirement; the ownership problem, which concerns the allocation of any rights that do subsist between the developer of the AI system, the operator, and the user who provides the prompt; and the training data question, which asks whether the ingestion of copyright works to train AI models constitutes infringement of reproduction rights.

The paper concludes that both India and Australia require legislative reform to address these challenges, and offers a comparative assessment of the reform options available to each jurisdiction.

Keywords

Copyright; Artificial Intelligence; AI-Generated Content; Authorship; Originality; Copyright Act 1957; Copyright Act 1968; India; Australia; Generative AI. I.

INTRODUCTION

Copyright law has always been compelled to adapt to new technologies of creation. The invention of photography required courts and legislators to determine whether a mechanical process of image capture could produce an authored work; the development of cinema raised questions about which of the many human contributors to a film's production held the rights in it; the arrival of the computer prompted statutory amendments to address works produced with computational assistance. Each of these adaptations was, in retrospect, an incremental adjustment to a framework whose foundational premise that creative works originate with human beings who deserve recognition and incentive for their creative labour remained undisturbed. Generative artificial intelligence threatens something more fundamental.

A large language model producing a novel, an image diffusion system generating artwork, or a music synthesis system composing a score does not merely assist a human author; it may replace the human creative act entirely, producing outputs whose aesthetic and commercial value is real but whose connection to any identifiable human author is tenuous or absent. The legal consequences of this displacement are profound and unresolved. If copyright exists to incentivise and reward human creativity, there is a principled case for concluding that AIgenerated works attract no copyright at all, leaving them in the public domain.

If, on the other hand, copyright serves the broader purpose of incentivising the production of valuable creative content for social benefit, there may be grounds for extending some form of protection to AI outputs, vesting it in the person or entity responsible for the system's development or deployment. Between these positions lies a spectrum of intermediate approaches that attempt to preserve some human creative contribution as the condition of copyright subsistence while acknowledging the transformative role of AI in the creative process. India and Australia offer a particularly instructive comparative context for examining these questions.

Both inherited the British colonial copyright framework and have developed their copyright law within the common law tradition; both have enacted relatively recent statutes the Copyright Act, 1957 in India and the Copyright Act, 1968 in Australia that are broadly similar in structure but differ in their specific treatment of computer-generated works. India's statute contains an express provision addressing computer-generated works, albeit one drafted before generative AI was a practical reality; Australia's statute has been interpreted by its courts in ways that raise significant questions about the copyright status of computationally generated outputs.

A comparison of these two jurisdictions illuminates both the shared challenges that the common law copyright tradition faces in the age of generative AI and the divergent paths that legislative and judicial response to those challenges might take. II. THEORETICAL FOUNDATIONS: AUTHORSHIP, ORIGINALITY, AND THE HUMAN CREATIVITY REQUIREMENT The concept of authorship has been central to copyright law since its origins in the Statute of Anne, and the association of authorship with human creative personality has been reinforced by both the natural rights and utilitarian traditions that underpin copyright doctrine in different legal cultures.

Copyright in the common law tradition does not require that a work demonstrate high aesthetic quality, but it does require that the work originate with the author that it be the product of the author's own intellectual effort rather than copied from another source. 1 The Supreme Court of India adopted a formulation of this requirement in Eastern Book Company v. D. B. Modak, holding that copyright subsistence demands a modicum of creativity not merely industrious collection or investment of effort, but the exercise of some element of intellectual judgment that reflects the author's personal intellectual choices. 2 The High Court of Australia took a similar direction in IceTV Pty Ltd. v. Nine Network Australia Pty Ltd.

, emphasising that copyright protection presupposes the exercise of independent intellectual effort by the author, and retreating from a purely sweat-of-the-brow approach that would protect investment of labour without requiring creative judgment. 3 The significance of these originality standards for AI-generated content lies in the nature of the creative process they presuppose. A work is original, under both the Indian and Australian approaches, because a human being made intellectual choices about it choices of selection, arrangement, expression, or form that could have been made differently and that reflect something of the author's personality or judgment. A generative AI system produces outputs 1University of London Press Ltd. v. University Tutorial Press Ltd.

, (1916) 2 Ch 601 (UK) (originating the principle that copyright requires only that the work originate with the author, not that it demonstrate aesthetic quality). 2Eastern Book Company v. D. B. Modak, (2008) 1 SCC 1 (Supreme Court of India) (adopting the 'modicum of creativity' standard for copyright subsistence, departing from the sweat of the brow doctrine). 3IceTV Pty Ltd. v. Nine Network Australia Pty Ltd. , (2009) 239 CLR 458 (High Court of Australia) (affirming that copyright subsistence in Australia requires some element of independent intellectual effort). through a process of statistical pattern matching applied to its training data; the outputs reflect the distribution of patterns in that data and the parameters set during training, not the exercise of intellectual choice in any sense that the doctrine of originality was designed to recognise.

The application of established originality standards to AI-generated works thus tends toward a conclusion that such works do not attract copyright, at least where the human contribution is limited to the provision of a prompt or the selection of a model. At the international level, WIPO's ongoing deliberations on IP and artificial intelligence have identified the authorship question as one of the most pressing issues facing the international intellectual property system, without reaching consensus on a recommended approach.

45 The divergence of national approaches with the United States Copyright Office consistently declining to register works without sufficient human authorship6 while the United Kingdom's Copyright, Designs and Patents Act provides expressly for computer-generated works7 illustrates the absence of any settled international norm on which national legislators can converge. III. THE INDIAN COPYRIGHT FRAMEWORK AND AI-GENERATED WORKS A. Statutory Definitions of Author and Computer-Generated Works The Copyright Act, 1957 defines 'author' in Section 2(d) by reference to specific categories of works: the author of a literary, dramatic, or musical work is the person who creates it; the author of an artistic work is the artist; the author of a cinematograph film is the producer; and the author of a sound recording is the producer.

8 Section 2(d)(vi), the provision most directly relevant to AI-generated works, addresses computer-generated works specifically: in relation to any literary, dramatic, musical, or artistic work which is computer-generated, the author shall 4WIPO, Revised Issues Paper on Intellectual Property and Artificial Intelligence, WIPO/IP/AI/2/GE/20/1 Rev. (Oct. 2020). 5WIPO, Draft Articles on IP and Genetic Resources, Traditional Knowledge and Folklore; see also WIPO Standing Committee on Copyright and Related Rights, AI and Copyright: Survey of National Approaches (2023). 6Thaler v. Vidal, 43 F. 4th 1207 (Fed. Cir. 2022) (holding that only human beings can be inventors for patent purposes; the reasoning has been applied by analogy in copyright debates); U. S. Copyright Office, Copyright and Artificial Intelligence, Part 1: Digital Replicas (2023).

7Copyright, Designs and Patents Act, 1988 (UK), s. 9(3) (providing that the author of a computer-generated work shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken). 8Copyright Act, 1957, No. 14, Acts of Parliament, 1957 (India), s. 2(d) (defining 'author' by reference to specific categories of works and the human beings responsible for their creation). be taken to be the person who causes the work to be created. 9 This provision has its antecedent in Section 9(3) of the UK Copyright, Designs and Patents Act, 1988, and reflects a legislative judgment that computer-generated works should attract copyright, with that copyright vesting in the human person responsible for causing the work to come into existence.

The provision was drafted at a time when 'computer-generated' referred primarily to outputs of deterministic programs operated by human programmers who directed the machine's actions at each step; its application to the outputs of large-scale generative AI systems trained on billions of parameters and operated through natural language prompts raises interpretive difficulties that the drafters did not contemplate. The central difficulty is the identification of the person who 'causes' a generative AI work to be created. The developer of the AI model made the fundamental choices about model architecture, training data, and optimisation objective that determine the character of the system's outputs. The operator who deploys the model configured the interface and set the parameters within which users interact with it.

The user who provides the prompt specifies the content of the particular output but typically exercises no control over the model's internal processes. Each of these persons exercises a form of causal influence over the final output, yet none of them corresponds neatly to the intuitive conception of the author as the person whose creative decisions shaped the work. B. Scope of Protection: Literary Works and Computer Programs The Copyright Act, 1957 extends copyright protection to literary works, which are defined in Section 2(o) to include computer programmes, tables, and compilations including computer databases. 10 Computer-generated outputs that take literary form text, code, structured data are therefore potentially protectable as literary works, subject to the originality requirement and the resolution of the authorship question.

The broader definitions of 'computer' in Section 2(ffc)11 and 'work' under the Act are sufficiently capacious to accommodate the outputs of generative AI systems in principle, but the absence of any provision addressing the originality 9Copyright Act, 1957 (India), s. 2(d)(vi) (defining the author of a computer-generated work as the person who causes the work to be created). 10Copyright Act, 1957 (India), s. 2(o) (defining 'literary work' to include computer programmes, tables, and compilations including computer databases). 11Copyright Act, 1957 (India), s. 2(ffc) (defining 'computer' broadly to include any electronic or similar device capable of storing, receiving, or processing information). or ownership of AI outputs beyond the basic computer-generated works provision leaves significant gaps that judicial interpretation has only partially filled.

The Delhi High Court's observations in Navigators Logistics Ltd. v. Kashif Qureshi addressed the requirement of human creative contribution in the context of computer-generated works, suggesting that even where Section 2(d)(vi) applies, some element of human intellectual engagement with the output is required to satisfy the originality standard. 12 This reading aligns the computer-generated works provision with the broader originality jurisprudence of the Supreme Court in Eastern Book Company, but it leaves unresolved the question of how much human engagement suffices and at which stage of the production process that engagement must occur. C. Moral Rights and AI-Generated Works A distinctive feature of the Indian copyright framework that bears directly on AI-generated content is the treatment of moral rights.

Section 57 of the Copyright Act, 1957 confers on the author of a work the right of integrity the right to object to any distortion, mutilation, or modification of the work that is prejudicial to the author's honour or reputation and the right of paternity, both of which survive the assignment of the economic rights in the work. 13 Where copyright in an AI-generated work is vested in a corporate developer or operator rather than in a natural person, the moral rights framework sits awkwardly: the rights are expressed in terms that presuppose a natural person author capable of having honour and reputation that the work reflects. The allocation of moral rights in AI-generated works is a question the Act does not address and that judicial interpretation has not yet resolved. IV. THE AUSTRALIAN COPYRIGHT FRAMEWORK AND AI-GENERATED WORKS A.

Authorship and Computer-Generated Works under the 1968 Act The Copyright Act, 1968 (Cth) does not contain an express provision equivalent to Section 2(d)(vi) of the Indian Act addressing computer-generated works by name. 14 The definition of 12Navigators Logistics Ltd. v. Kashif Qureshi, 2018 SCC OnLine Del 9454 (Delhi High Court) (discussing the scope of copyright in computer-generated outputs and the requirement of human creative contribution). 13Copyright Act, 1957 (India), s. 57 (moral rights of authors, including the right of integrity and the right of paternity, which survive assignment of copyright).

14Copyright Act, 1914 (Australia); Copyright Act, 1968 (Cth) (Australia) (the primary statute governing 'author' in the Act is not comprehensively enumerated in the manner of the Indian provision; rather, authorship is understood through the general principles of the common law tradition, in which the author is the natural person whose intellectual effort produced the work. The Act's definition of 'artistic work' and related categories encompasses works produced with computer assistance, but the question of whether a work produced entirely by a computer system without sufficient human creative input can attract copyright has been addressed primarily through judicial rather than legislative elaboration. The Full Federal Court of Australia's decision in Acohs Pty Ltd. v. Ucorp Pty Ltd.

addressed this question in the context of a computer-generated database and held that material generated by a computer program without meaningful human authorial contribution at the point of creation does not attract copyright protection as an original work. 15 The Court's reasoning proceeded from the principle affirmed in IceTV that copyright requires independent intellectual effort attributable to a human author; where the relevant creative choices are made by the computer program rather than by a human being, the condition of copyright subsistence is not satisfied. The implication for AI-generated content is significant: under the current Australian framework, works produced by generative AI without sufficient human creative contribution at the point of generation are likely to fall in the public domain. B.

The Thaler Litigation and Its Implications Australia's engagement with the question of AI and intellectual property rights has been most visible in the Thaler litigation, which concerned the cognate question of whether an AI system could be named as an inventor for patent purposes. At first instance, the Federal Court held that the Patents Act admitted of an AI inventor; on appeal, the Full Federal Court reversed, holding that inventors within the meaning of the Act must be natural persons, a conclusion that the High Court declined to disturb. 16 While the Thaler litigation concerned patents rather than copyright, its reasoning about the human requirement for intellectual property protection is copyright in Australia). 15Acohs Pty Ltd. v. Ucorp Pty Ltd.

, (2012) 201 FCR 173 (Full Federal Court of Australia) (holding that a work generated by a computer program without sufficient human authorial contribution may not attract copyright protection). 16Thaler v. Commissioner of Patents, (2021) 92 IPR 1 (Fed. Ct. Aust. ) (holding that an AI system could be named as an inventor for patent purposes); reversed on appeal: Commissioner of Patents v. Thaler, (2022) 295 FCR 224 (Full Federal Court of Australia) (overturning the primary decision and holding that inventors must be human persons). directly relevant to the copyright context and reflects the Australian judiciary's reluctance to extend IP rights to non-human creators without express legislative authorisation.

The Australian Law Reform Commission's examination of copyright and the digital economy, published in 2013, addressed a range of questions about the adequacy of the Copyright Act for the digital context but predated the emergence of generative AI as a commercial phenomenon. 17 More recent engagement with the AI and copyright question has come through the Australian Government's broader AI governance consultations, which have addressed the copyright dimensions of AI training, AI-generated content, and the liability of AI developers for infringement by their systems. 18 As of mid-2026, Australia has not enacted legislation specifically addressing copyright in AI-generated works, and the question remains to be resolved through the application of existing doctrine or through legislative reform. C.

Fair Dealing and the Training Data Question in Australia The Copyright Act, 1968 contains a set of fair dealing exceptions for research, study, criticism, review, and news reporting that are broadly analogous to the fair dealing provisions of the Indian Act, though Australia has not adopted the open-ended fair use doctrine of the United States. 19 The Australian Productivity Commission's 2016 review of IP arrangements recommended the introduction of a broader fair use exception, and the ALRC made a similar recommendation in 2013, but neither recommendation has been enacted.

20 The question of whether the ingestion of copyright works to train a generative AI model constitutes fair dealing or whether it requires a new statutory exception is one of the most practically significant unresolved questions in Australian copyright law, with implications for both AI developers and the creative industries whose works have been used without licence or compensation. 17Attorney-General's Department, Australian Government, Copyright and the Digital Economy: Final Report (ALRC Report 122) (2013) (Australian Law Reform Commission examination of copyright in the digital context). 18Department of Industry, Science and Resources, Australian Government, Safe and Responsible AI in Australia: A Discussion Paper (2023); Australian Government, Voluntary AI Safety Standard (2024). 19Copyright Act, 1968 (Cth) (Australia), ss.

40-42 (fair dealing exceptions for research, study, criticism, review, and reporting news); compare 17 U. S. C. s. 107 (fair use doctrine, a broader and more flexible exception applied in the United States). 20Parliamentary Standing Committee on Communications and the Arts, Parliament of Australia, Inquiry into the Australian Music Industry (2020); Australian Government Productivity Commission, Report on IP Arrangements (2016).

V. COMMON

CHALLENGES

AUTHORSHIP GAP, OWNERSHIP, AND TRAINING DATA A. The Authorship Gap The most fundamental challenge for both the Indian and Australian copyright frameworks is what may be described as the authorship gap: the disjunction between the legal requirement of human creative authorship and the practical reality of AI-generated content whose aesthetic and commercial value does not depend on any identifiable human creative act at the point of generation. Both jurisdictions, in their different ways, have arrived at a position in which copyright subsistence requires some human creative contribution, but neither has defined with precision what contribution suffices in the context of generative AI. The authorship gap has three practical manifestations.

First, works produced entirely by AI without human creative input where the user provides only a minimal or formulaic prompt almost certainly fall outside copyright protection under both the Indian and Australian frameworks as currently interpreted. The commercial consequences of this conclusion are significant: the developer who has invested in building the generative AI system, and the operator who has deployed it, are unable to prevent competitors from reproducing the outputs freely. Second, works produced through intensive human-AI collaboration where the user exercises substantial creative judgment in selecting, refining, and composing AI-generated elements may attract copyright protection on the basis of the human contribution, but the line between protectable human-AI collaboration and unprotectable AI generation is not clearly drawn in either jurisdiction.

Third, the outputs of AI systems trained on and closely imitating the style of particular human authors raise questions about the relationship between the human training corpus and the AI output that copyright doctrine has no established framework to address. B. The Ownership Problem Even where copyright is held to subsist in an AI-generated work, the allocation of ownership between the developer of the AI system, the operator, and the user raises questions that neither the Indian nor the Australian statute directly resolves. The Indian provision vesting copyright in the person who 'causes' the work to be created identifies the relevant criterion but leaves its application to context-specific interpretation.

The Australian framework offers no equivalent statutory guidance, leaving the question entirely to common law principles of employment, commission, and equitable ownership that were developed without AI in contemplation. 21 The commercial arrangements through which generative AI systems are typically deployed add further complexity. The terms of service governing user access to AI platforms routinely allocate intellectual property rights in outputs between the platform and the user in ways that may or may not correspond to the statutory allocation under applicable copyright law, and the enforceability of these contractual allocations against the background of statutory rules that cannot be contracted out of remains an open question in both jurisdictions. C.

The Training Data Question The ingestion of copyright works to train generative AI models is the dimension of the AI and copyright relationship that has attracted the most litigation internationally, with class actions by visual artists and stock photography agencies against AI developers filed in the United States raising questions whose resolution will have significant persuasive influence on how Indian and Australian courts and legislators address the same issue. 22 23 Under both the Indian and Australian copyright frameworks, the reproduction of a substantial part of a copyright work without authorisation constitutes infringement, and the fair dealing exceptions are construed narrowly in both jurisdictions.

24 The argument that AI training constitutes reproduction of a substantial part of ingested works because the model's parameters encode statistical representations of those works that can, under some circumstances, be recovered in the model's outputs is plausible but contested. The counter-argument, that training is a transformative process that does not reproduce any expression from the ingested works in the model itself, draws on the transformative use reasoning developed in United States fair use jurisprudence but has no direct doctrinal foundation in the Indian or Australian fair dealing frameworks. 21Copyright Act, 1968 (Cth) (Australia), ss. 31-35 (rights comprised in copyright and ownership of copyright in various categories of works). 22Andersen v. Stability AI Ltd. , Case No. 3:23-cv-00201 (N. D. Cal. , filed Jan.

13, 2023) (class action by visual artists alleging copyright infringement through the use of their works to train AI image generation systems). 23Getty Images (US) Inc. v. Stability AI Ltd. , Case No. 1:23-cv-00135 (D. Del. , filed Feb. 3, 2023) (copyright and trade mark infringement claims arising from use of licensed image library for AI training). 24Copyright Act, 1957 (India), s. 52 (fair dealing provisions permitting use of copyright works for purposes including research, private study, criticism, review, and reporting of current events).

The absence of a text and data mining exception in either the Indian or Australian copyright statute is a significant lacuna. The European Union's Copyright in the Digital Single Market Directive introduced a mandatory text and data mining exception for research purposes and a qualified exception for commercial purposes, providing a model that both India and Australia have been discussing but neither has enacted. Japan's broad text and data mining exception has attracted commentary as the most permissive approach globally, and has been pointed to by AI developers as the appropriate model for jurisdictions that wish to foster AI innovation.

The policy choice between protecting the interests of creators whose works are used in training and enabling AI developers to train on the cultural commons without licensing obligations is one of the defining copyright policy questions of the current decade. VI.

COMPARATIVE

ANALYSIS

AND LEGISLATIVE REFORM OPTIONS A. Strengths and Weaknesses of Each Framework The Indian framework's express provision for computer-generated works represents a structural advantage over the Australian position: it provides a statutory basis for copyright in AIgenerated works that the Australian Act lacks, and the allocation of that copyright to the person who causes the work to be created provides at least a starting point for ownership analysis. However, the provision's drafting history reflecting a conception of computer generation as a human-directed deterministic process means that its application to genuinely autonomous generative AI involves significant interpretive extension that the courts have not yet comprehensively undertaken.

The Australian framework's reliance on common law originality principles produces a more principled but less certain result: the Acohs decision points toward a conclusion that AIgenerated works without sufficient human creative contribution attract no copyright, which is at least doctrinally coherent, even if it leaves a commercially significant category of valuable content unprotected. The Australian Government's active consultation on AI governance and the productivity commission's prior recommendations on fair use create a legislative environment more conducive to comprehensive reform than India's, where the equivalent policy discussions are at an earlier stage. 25 25Ministry of Electronics and Information Technology, Government of India, National Strategy for Artificial Intelligence (NITI Aayog, 2018); Ministry of Electronics and Information Technology, Government of India, B.

Reform Options Common to Both Jurisdictions Several reform options present themselves as relevant to both jurisdictions. First, both could introduce express provisions clarifying the copyright status of AI-generated works and the conditions under which copyright subsists in them for example, by requiring a minimum threshold of human creative contribution, defined by reference to the creative choices made in designing the prompt, selecting or curating outputs, or configuring the AI system. Second, both could address the ownership question directly, providing a default rule for the allocation of copyright in AI outputs between developers, operators, and users in the absence of contractual agreement.

Third, both could introduce or expand text and data mining exceptions to provide a clearer legal basis for the use of copyright works in AI training, potentially combined with remuneration mechanisms to compensate rights holders whose works are ingested at scale. The WIPO processes on IP and AI have identified these three areas authorship and ownership of AI outputs, the training data question, and liability for AI-generated infringement as the priority issues for international norm-setting, and both India and Australia have been active participants in those processes. India's National IPR Policy frames intellectual property as a driver of innovation and creative industry development a framing that provides policy support for extending some form of protection to commercially valuable AI-generated content.

Australia's approach to AI governance has emphasised safety and responsibility a framing that may give greater weight to the interests of human creators whose works and livelihoods are affected by AI-generated competition. VII.

CONCLUSION

The comparative analysis of the Indian and Australian copyright frameworks reveals that both jurisdictions are confronting the same fundamental challenge the application of doctrines built on human authorship to works produced by machines from positions that are more similar in their difficulties than their statutory differences might suggest. India's express provision for computer-generated works provides a textual hook that Australia's Act lacks, but both frameworks ultimately depend on an originality standard that requires human creative contribution and an authorship concept that presupposes a natural person whose intellectual effort produced the work. Generative AI destabilises both of these requirements Principles for Responsible AI (2021). simultaneously, and neither statute, as currently drafted and interpreted, provides a fully adequate response to the challenge.

The three challenges identified in this paper the authorship gap, the ownership problem, and the training data question are interconnected rather than discrete, and their resolution will require legislative choices about the underlying purposes of copyright law and its relationship to the AI economy. If copyright exists primarily to incentivise and reward human creativity, there is a strong case for refusing protection to AI-generated works and for ensuring that the training of AI systems on human-authored works is subject to licensing requirements that compensate the creators whose works have made the systems possible.

If copyright also serves the purpose of ensuring that valuable cultural content is produced and made available, there may be grounds for extending some form of sui generis or adapted protection to AI outputs while addressing the training data question through a combination of exceptions and remuneration schemes. What is clear is that the status quo in which neither jurisdiction has enacted legislation directly addressing these questions and both rely on analogical extension of pre-AI doctrines is unsatisfactory.

Courts in both India and Australia will increasingly be asked to resolve disputes that their existing frameworks are ill-equipped to address, and the answers they give in the absence of legislative guidance will be shaped as much by the incidental features of the cases brought before them as by any coherent policy vision for the relationship between copyright and artificial intelligence. Timely and thoughtful legislative reform, informed by the comparative experience of other jurisdictions and the emerging international consensus at WIPO, is the most promising path toward a copyright framework adequate to the creative and economic realities of the age of generative AI.

BIBLIOGRAPHY

A. Legislation Copyright Act, 1957, No. 14, Acts of Parliament, 1957 (India). Copyright Act, 1968 (Cth) (Australia). Copyright, Designs and Patents Act, 1988 (UK). B. Cases Acohs Pty Ltd. v. Ucorp Pty Ltd. , (2012) 201 FCR 173 (Full Federal Court of Australia). Andersen v. Stability AI Ltd. , Case No. 3:23-cv-00201 (N. D. Cal. , filed Jan. 13, 2023). Attorney-General v. Guardian Newspapers Ltd. (No. 2), (1990) 1 AC 109 (UK) (on the public domain). Burrow-Giles Lithographic Co. v. Sarony, 111 U. S. 53 (1884). Commissioner of Patents v. Thaler, (2022) 295 FCR 224 (Full Federal Court of Australia). Eastern Book Company v. D. B. Modak, (2008) 1 SCC 1 (Supreme Court of India). Getty Images (US) Inc. v. Stability AI Ltd. , Case No. 1:23-cv-00135 (D. Del. , filed Feb. 3, 2023). IceTV Pty Ltd. v. Nine Network Australia Pty Ltd. , (2009) 239 CLR 458 (High Court of Australia).

Navigators Logistics Ltd. v. Kashif Qureshi, 2018 SCC OnLine Del 9454 (Delhi High Court). Thaler v. Commissioner of Patents, (2021) 92 IPR 1 (Federal Court of Australia). Thaler v. Vidal, 43 F. 4th 1207 (Fed. Cir. 2022). University of London Press Ltd. v. University Tutorial Press Ltd. , (1916) 2 Ch 601 (UK). C. Government Documents, Reports, and Policy Papers Arts Law Centre of Australia, Artificial Intelligence and Copyright: Key Issues for Australian Creators (2023). Attorney-General's Department, Australian Government, Copyright and the Digital Economy: Final Report (ALRC Report 122) (2013). Australian Government, Voluntary AI Safety Standard (2024). Department of Industry, Science and Resources, Australian Government, Safe and Responsible AI in Australia: A Discussion Paper (2023).

Ministry of Commerce and Industry, Government of India, National Intellectual Property Rights Policy (2016). Ministry of Electronics and Information Technology, Government of India, National Strategy for Artificial Intelligence (NITI Aayog, 2018). Ministry of Electronics and Information Technology, Government of India, Principles for Responsible AI (2021). Parliamentary Standing Committee on Communications and the Arts, Parliament of Australia, Inquiry into the Australian Music Industry (2020). Productivity Commission, Australian Government, Report on IP Arrangements (2016). U. S. Copyright Office, Copyright and Artificial Intelligence, Part 1: Digital Replicas (2023). D. International Instruments and Organisational Documents WIPO, Revised Issues Paper on Intellectual Property and Artificial Intelligence, WIPO/IP/AI/2/GE/20/1 Rev. (Oct. 2020).

WIPO Standing Committee on Copyright and Related Rights, AI and Copyright: Survey of National Approaches (2023).