From Human Hands To Machine Minds - The Authorship Dilemma

Author: LawSphereX

Category: General Law

Published: 2026-08-23T19:53:52.790Z

Updated: 2026-08-30T07:05:15.851Z

FROM HUMAN HANDS TO MACHINE MINDS: THE

AUTHORSHIP DILEMMA Anshika Kumari, National Law University, Odisha Jomjar Padu, National Law University, Odisha I. Introduction The rise of generative AI has changed creativity in different ways which refuses long term held notions relating to authorship and originality. Copyright law since long assumed that creativity in humans is the product of skill, labour and imagination. Machine learning systems nowadays, trained on huge data sets are writing poetry, composing music, creating visual art, and even academic writing with little or no ongoing human intervention. 1 This movement presents a pressing question: who, if anyone, can justifiably lay claim to such pieces of work? The argument over AI authorship is complicated. Whether human input is enough if it's refining prompts or tweaking algorithms is always a question of meeting criteria for originality.

There are also rival claims between users and owners of AI over control and ownership. On an international scale, regulatory fragmentation makes the issue even more complicated since legal systems are at variance on what to categorize AI-created works. Apart from law, there's also the issue of human creativity as a whole: if content created by AI becomes predominant, will this kill the incentive for people to create? 2 This essay argues that while copyright must remain grounded in human authorship, it should also evolve to acknowledge AI’s transformative role through policies that balance innovation, fairness, and cultural enrichment. Authorship has been based on philosophy and Salmond also defined that authorship is granted to legally recognised person and not biologically which anyone can be author if law recognises it as person.

There have been theories which shaped copyright law namely Labour theory given

1 admin, ‘Beyond Human Authorship: The Legal Dilemma of AI-Generated Content’ (The IP Press, 12 July

2025) accessed 30 September 2025. 2 Peidong Mei and others, ‘If ChatGPT Can Do It, Where Is My Creativity? Generative AI Boosts Performance but Diminishes Experience in Creative Writing’ (2025) 4 Computers in Human Behavior: Artificial Humans 100140. by Lockean justifying rights through skills and efforts, then Personality theory given by Hegelian which treated work as extension of will and lastly Utilitarian theory which viewed authorship as a way to promote cultural progress, and together these shaped he copyright law around labour, personality and societal benefits.3 II. The Legal Landscape – National and Global Approaches

1. The Indian Position

India has cautiously engaged with the question of AI-generated authorship. The Copyright Act, 1957, under Section 2(d)(vi)4 (introduced by the 1994 amendment), provides that in the case of computer-generated works, the “author” is “the person who causes the work to be created. ” This provision appeared to anticipate a future where non-human agents could generate works. However, Indian courts have insisted that authorship must be vested in natural persons. In Rupendra Kashyap v. Jiwan Publishing House Pvt. Ltd. 5, the Delhi High Court categorically rejected the claim of an artificial entity to copyright, affirming that the statutory term “person” should not be extended to machines. Similarly, the Supreme Court in Som Prakash Rekhi v. Union of India6 held that “personality” under law implies the capacity to bear rights and duties something AI lacks. Furthermore, in Eastern Book Company v.

D. B. Modak7, the Supreme Court shifted India from the “sweat of the brow” standard to the “modicum of creativity” test, aligning with the U. S. approach. This doctrine makes copyright protection conditional on human intellectual contribution, raising doubts about the protectability of works produced autonomously by AI, where human input is minimal.

2. United States

The U.S. has adopted a strict human-authorship approach stated under the Copyright Act, 1976

3 Jessica Meindertsma, ‘Theories of Copyright | Copyright Corner’

accessed 30 September 2025.

4 Copyright Act, 1957, s 2(d)(vi) (India).

5 1996 (38) DRJ 81

6 (1981) 1 SCC 449

7 2008 (36) PTC 1 (SC)

that originality and creativeness should must come from human mind. In Thaler v. USPTO and Thaler v. Vidal, the U.S. courts refused to recognise an AI System named DABUS as an inventor or author. Similarly the U.S. Copyright office also clarified that any AI generated work if lacks human authorship cannot be granted copyright.8 However, nuanced cases exist. In Zarya of the Dawn, the USCO allowed copyright for a graphic novel containing AI-generated images, but only for the human arrangement, sequencing, and modification of those images. This difference shows a practical approach that AI outputs lacks protection but any human creation of those outputs can be protected.9

3. European Union

The EU has been proactive in regulating AI’s impact on IP. The EU AI Act and GDPR establish transparency obligations and safeguard against misuse of data in AI systems. In copyright law, however, the EU remains human-centric. A 2020 European Commission report rejected granting AI systems authorship or inventorship. Courts have reinforced this stance: for example, the Municipal Court of Prague held that an AI-generated image lacked protectable originality, as it did not arise from human creative activity.10 Nonetheless, policy debates within the EU suggest openness to new legal categories or sui generis frameworks to protect AI-generated outputs, particularly in patent contexts.

4. WIPO (World Intellectual Property Organization)

The World Intellectual Property Organization (WIPO) has positioned itself as the global convener on AI and IP. In its 2019 study, WIPO reported nearly 340,000 AI-related inventions had been patented worldwide. Former Director General Francis Gurry suggested that, from an economic perspective, there may be no reason to exclude AI from IP protection, though

8 Ji Mao, ‘Revisiting AI Inventorship in Thaler v. Vidal | Insights | Holland & Knight’

accessed 30 September 2025. 9 Anirudh Jonnavithula, ‘Zarya of the Dawn: How AI Is Changing the Landscape of Copyright Protection’ (Harvard Journal of Law & Technology, 6 March 2023) accessed 30 September 2025. 10 Adnan Masood, ‘Intellectual Property Rights and AI-Generated Content — Issues in Human Authorship, Fair Use Doctrine, and Output Liability | by Adnan Masood, PhD. | Medium’ (Medium, 4 April 2025) accessed 30 September 2025. moral and legal complexities remain. WIPO’s ongoing “Conversation on AI and IP” explores whether a sui generis regime with limited rights, mandatory registration, and clear disclosure of AI involvement could reconcile innovation incentives with fairness. 11 However, no international consensus has yet been reached. III. The Authorship Dilemma – Human Input vs.

Machine Autonomy A central challenge in the authorship debate is determining when human contribution crosses the threshold from prompting to authorship, and when machine outputs remain unownable. Copyright has long anchored originality in human input, yet generative AI increasingly produces content with little or no human involvement. The key question becomes: how much human involvement is required for a work to be considered “authored” in law? 12 Courts and policymakers distinguish between AI-assisted and AI-autonomous works. In Thaler v. USPTO, the U. S. court held that merely initiating an AI process does not make one an “author”; continuous human evaluation and alteration are required. Similarly, the U. S. Copyright Office clarified that entering a simple prompt does not suffice authorship demands “creative control” through selection, sequencing, or modification.

13 India’s 2025 draft CRI Guidelines echo this stance, vesting rights in humans who meaningfully direct AI outputs. Prompt engineering alone rarely meets the “modicum of creativity” test from D. B. Modak, whereas iterative refinement or artistic arrangement may justify protection. 14 Originality is still the foundation of copyright. Detractors say that AI simply recombines patterns of data, but advocates posit that novel outputs might meet originality even without human intent. The Zarya of the Dawn case shows a middle ground: AI images were not protected, but the human arrangement made into a graphic novel was protected. 15 Aside from originality, the control-versus-ownership issue continues. Copyright has historically rewarded creativity rather than investment. Anchoring rights in developers threatens monopolisation and

11 Yiyang Mei, ‘Prompting the E-Brushes

Users as Authors in Generative AI’ (Social Science Research Network, 6 March 2024) accessed 30 September 2025. 12 Mackenzie Caldwell, ‘What Is an “Author”?-Copyright Authorship of AI Art Through a Philosophical Lens’ (2023) 61 Houston Law Review 411.

13 Mao (n 8).

14 2008 (36) PTC 1 (SC)

15 Jonnavithula (n 9).

excludes true human-AI collaborations. Ultimately, this dilemma points to a more profound question: is creativity human or can outputs determine it? IV. Comparative Insights and Emerging Policy Models The authorship challenge is not limited to India; it has been a worldwide enigma, with jurisdictions trying different models for AI-created works. European Union The EU takes a conservative, human-centered approach. The European Commission in 2020 agreed that AI would not be accorded rights like authorship or inventorship. Reforms instead are aimed at modifying existing IP regimes and establishing new categories where necessary. The EU AI Act and GDPR place duties on transparency and use of data, influencing how AI works in creative industries.

Courts are still traditional: the Municipal Court of Prague decided an AI-produced image cannot be a "work," since it did not involve any creative input by a human. 16 United States The U. S. Copyright Office repeatedly denies AI as an author, reiterating copyright safeguards human creativity. In Thaler v. Vidal, courts asserted AI cannot be named as an inventor, echoing copyright law. 17 Yet Zarya of the Dawn reflects a partial-protection model: while AI images were denied copyright, the author was recognised for sequencing them into a narrative. 18 U. S. law thus ties protection to the “modicum of creativity” standard, granting rights only where human judgment is evident. WIPO The World Intellectual Property Organization facilitates international dialogue. Its 2019 study noted 340,000 AI-related patents worldwide but highlighted the absence of consensus on authorship.

WIPO discussions have floated a sui generis regime for AI-generated works, 16 Nandita Saikia, ‘Artificial Intelligence: An Opportunity to Recast Copyright Law’ (Social Science Research Network, 14 May 2025) accessed 30 September 2025.

17 Mao (n 8).

18 Jonnavithula (n 9).

granting limited rights tied to registration, disclosure, and human oversight. Emerging Policy Models Globally, four broad policy models are in the process of emerging. The first is the Human- Centric Model, in which rights are limited to human-authored contributions, reflecting the EU and U. S. approach. The second is the Control/Investment Model, in which ownership is linked with the developer or operator of the AI system, relating to patent law and the DABUS debate. The third is the Sui Generis Protection Model, with limited rights for AI-created works, subject to registration, disclosure, and evident marking of human participation, as evident in WIPO deliberations and Japan's experiments with legislation.

Lastly, the Public Domain Approach considers AI output as unowned and available to everyone, promoting openness and innovation but potentially deterring human-AI collaborative creativity. 19 These international experiments mirror the lack of consensus and regulatory fragmentation risk, an issue which is acuter still in the light of counter-arguments to a human-centred strategy. V. Counter-Arguments and Rebuttals No discussion of the authorship dilemma is possible without regard to the competing views which militate against a human-centred strategy. Copyright controversies has given rise to some models which looks appealing on the surface but flawed when closely looked, one is to give authorship to those who created or own the AI system. The supporters say that time, knowledge and money have been spend by the developers without which no output would have been ever possible.

They tend to equate this with principles of law of patents or the “work for hire” doctrine, wherein rights track the party in control and expending the cost. This argument fails, however, the fundamental precepts of copyright. Copyright exists to protect creativity, not capital or work. Developers are already protected in their software code and, where suitable, can patent their technical innovation. Expanding authorship to all works produced by AI would be akin to a “double dipping”, conferring sole control over a wide range of creative spaces with no human expression underlying. 20 19 Pehr-Johan Norbäck and Lars Persson, ‘Why Generative AI Can Make Creative Destruction More Creative but Less Destructive’ (2024) 63 Small Business Economics 349.

20 admin (n 1).

Indian courts have time and again emphasized this distinction. In Rupendra Kashyap v. Jiwan Publishing House, 21it was unmistakably stated that originality lies in human contribution. Similarly, in Eastern Book Co. v. D. B. Modak22, the Supreme Court elucidated that effort or investment is not enough; a spark of human imagination, no matter how small, is a requirement. These precedents establish that even though developers might enable production, facilitation alone cannot be authorship in the eyes of the law. Another school of thought assigns responsibility to prompt engineers as the actual authors. Iterative refinement of prompts and judicious editing arguably display a kind of craft comparable to that of a film director or photograph composer.

This view found partial support in the United States in the Zarya of the Dawn case, where copyright was granted not for images created by AI but for human authorship of presenting them as a graphic novel. But to equate every prompt with authorship is to threaten trivialisation of originality. Instruction in one line cannot match genuine artistic judgment. A fairer position is India's forthcoming “creative control” test, which judges authorship on substantial and perceptible human contribution instead of mechanical stimulation. 23 A third option envisions sui generis protection of AI works of creativity. Bounded rights, with transparency and bounded time durations, would be investor-friendly without warping copyright. The Japan AI Innovation Act of 2025 is a pilot case. 24 Sui generis regimes risk fragmenting global IP law and creating enforcement issues across borders.

Overlaps with current database and computer software rights are unclear, too. It is only ancillary protection and surrogate to human-generated copyright. Lastly, there are those who advocate that all works composed by AI be made public domain. It maximizes use and abolishes monopolies in non-human authors. Although pleasing, it suppresses incentive to invest in socially beneficial uses of AI and denies moral rights such as attribution rights which reside at the heart of the Hegelian and Kantian cases in favor of copyright. 25

21 1996 (38) DRJ 81

22 2008 (36) PTC 1 (SC)

23 Jonnavithula (n 9).

24 ‘Understanding Japan’s AI Promotion Act: An “Innovation-First” Blueprint for AI Regulation’

(https://fpf.org/) accessed 30 September 2025.

25 Mei and others (n 2).

All these arguments on the contrary in combination speak about the variety of potential reactions. A developer-driven model incentivizes capital more than imagination; a promptdriven model incentivizes trivialisation; a sui generis solution incentivizes fragmentation and a public-domain solution incentivizes disincentivising of innovation. The only way is to adopt a hybrid approach, maintaining copyrights efficiency by assigning authorship to human creativity and include responsibility requirements through transparency and give judicial mechanisms for AI-related works. This can increase the scale of innovation and fairness by recognising worth of human creativity. VI. Conclusion The AI authorship debate shows that copyright law is now on the critical crossroads.

The traditional grounds of creativity have been upset by generative AI, but the fundamental principle is clear: copyright has to be meant for human ingenuity and not mechanical processes. The jurisdictions of the world, whether India, the US, or the European Union, keep reiterating that originality has to involve some degree of human judgment, skill, and imagination. Courts have consistently explained that investment of money or just facilitation is not enough, and such differentiation keeps copyright grounded in its philosophical origin in labour, personality, and societal advancement. At the same time, the transformative potential of AI cannot be ignored. Policy models ranging from human-centric to public-domain approaches demonstrate both the diversity of thought and the difficulty of achieving global consensus.

What emerges most persuasively is the need for a balanced, hybrid approach—one that safeguards human authorship while accommodating the technological realities of AI. Such a system should recognize true creative guidance, institute transparency, and avoid monopolization of cultural spaces. By reaffirming the place of human creativity while conforming to technological evolution, copyright can still foster innovation, safeguard dignity, and maintain the dynamism of the public domain.