Ownership Of AI-Generated Works - Rethinking Authorship And Inventorship In IP Law

Author: LawSphereX

Category: Intellectual Property

Published: 2026-08-23T20:42:36.137716+00:00

Updated: 2026-08-26T17:10:14.162+00:00

OWNERSHIP OF AI-GENERATED WORKS: RETHINKING

AUTHORSHIP AND INVENTORSHIP IN IP LAW

Nitya Pasari, LLB (Hons.), Jindal Global Law School, O.P. Jindal Global University

ABSTRACT

The shift to autonomous and generative systems from traditional AI, is challenging the doctrinal and conceptual foundations of Intellectual Property Law. The essence of the law has always been human creativity, intention for creation, authorship, ownership, and inventorship, the current legal frameworks are struggling to accommodate AI-generated works within this framework with minimal human intervention. This paper critically examines the doctrinal limitations of Indian IP law along with some comparison with international precedents, with major focus on the issues pertaining to authorship under copyright law and inventorship under patent law. The paper also explores options of ownership models along with their pros and cons including human-centric, AI-as-author, public domain and sui generis approaches.

Using the guidance of judicial precedents and scholarly articles, this paper highlights the incompatibility between human-centric legal doctrines and machine-generated outputs. Further towards the end, the paper identifies key policy challenges and shows that there is an urgent need for adaptive legal reform which balances innovations with AI-generated products along with doctrinal coherence, particularly in jurisdictions like India where statutory and policy gaps persist.

INTRODUCTION

Intellectual Property law deals with laws to protect and enforce rights of the creators and owners of inventions, writing, music, designs and other works, known as the “intellectual property”1. Concepts such as authorship, inventorship, originality, creativity, and moral rights are traditionally rooted in the assumption of human intellect, intention, and agency. However, the rise of advanced and generative AI, has begun to destabilize these foundational assumptions. Contemporary generative AI systems operate with increasing independence and minimal human intervention, unlike traditional form of AI, where human intervention played a major tool and which functioned as assisting humans in decision-making. They are trained through vast datasets and driven by complex algorithms.

This shift from traditional AI to contemporary generative AI raised an important question: “Can a machine be considered an author or inventor? ” If not, who owns the output generated by such systems? And how should existing IP frameworks respond to this technological disruption, which causes chaos? In the Indian context, these questions are particularly pressing as the statutory provisions continue to reflect a strongly human-centric approach. Leaving limited scope for AI-generated contributions. While global developments- such as the DABUS litigation- have intensified the debate, Indian courts has not yet confronted this issue directly. This paper seeks to examine the growing tension between technological advancement and doctrinal rigidity in IP law. It analyses the foundational concepts of authorship in copyright law and inventorship in patent law under Indian law.

The paper further explores competing ownership models, and reads Indian judicial precedents with respect to AI-generated works. On identifying key challenges and policy concerns, the paper conclusively argues for reexamination of existing legal frameworks to fill the gap wit respect to machine-generated innovation.

CONCEPTUAL FOUNDATION UNDER INDIAN IP LAW

Authorship under Copyright Law 1 Georgetown Law, ‘Intellectual Property Law’ https://www.law.georgetown.edu/your-life-career/careerexploration-professional-development/for-jd-students/explore-legal-careers/practice-areas/intellectual-propertylaw/ accessed 27 April 2026.

The Copyright Act, 19572, in India governs the protection of original creative works, where Section 2(d), has defined Authorship as the act of creating the work. The Act specifies a modern addition with respect to computer-generated works, where a person is credited who causes the work to be created, acknowledging technological advancements. However, this reveals a significant grey area because the provision assumes 3 major things: Ø Human Control; Ø Human Intention; and Ø Human Causation. These assumptions become problematic, in the context of autonomous AI. When an output is generated through any AI system independently, identifying “the person causing the work”, as stated in the statute becomes a legal fiction.

In traditional AI-generated works, human input and the above assumptions were central, but now in the recent time of advances AI system, there might be minimal human involvement, raising doubts of this falling within the statutory definition. • Inventorship under Patent Law In simpler terms, the inventors are the ones who developed the invention- the creative minds behind the patent3. The Patents Act, 19704, requires the inventor to be a natural person, which leaves no room for non-human inventors, reinforcing the human-centric nature of Indian patent law- and excluding AI systems as inventors. However, this has been globally contested “whether AI system can be recognised as a true and first inventor”- specifically in the DABUS litigation initiated by Dr. Stephen Thaler. This will be discussed further more in the paper.

2 Copyright Act 1957 (India).

3 Michael K Henry, ‘Patent Ownership vs Inventorship: Who Really Controls the Rights to a Patent?’ (Henry Patent Law Firm, 23 October 2024) https://henry.law/blog/patent-ownership-vs-inventorship/ accessed 27 April 2026.

4 Patents Act 1970 (India).

LITERATURE REVIEW

The intersection of AI and IP law are certainly on a long journey where reconciliation will take time. It has attracted significant scholarly attention in recent years, particularly with respect to the challenge posed by the advanced AI system and the ownership and authorship of the work generated through it. AI- through its decision-making and content-generation capabilities is pushing the boundaries of what IP was originally designed to protect, making those boundaries increasingly ambiguous5. One of the major debates which lies at the hearts of the scholars is “Whether AI-generated content can be considered original under copyright law”6? Courts and legal experts say that originality (as mentioned in Section 13 of the Copyrights Act7), comes from human imagination, labour and intent, which AI does not have.

Another unfortunate obstacle in AIgenerated content, is finding out who is accountable for copyright infringement8. Recent studies such as of Ahlawat and Chopra9 say that the current legal framework is inadequate to address such issues. NITI Aayog published the landmark report: “National Strategy for Artificial Intelligence- #AIforAll” (2018)10, where clearly, they have focused on AI-driven innovation and economic growth but it did not directly address IP ownership, which reveals a policy gap. This report demonstrates that AI must remain human-controlled. It supports our argument that: Indian policy still views AI as a toll, not an autonomous actor. Therefore, authorship must remain human-centric under the current policy thinking of India.

Overall, the literature reveals three major trends: 5 Bagath Manish, ‘Indian Perspective of Intellectual Property for AI-Created Works’ (2025) 3(2) Trends in Intellectual Property Research 44 https://www. researchgate. net/publication/395282362_Indian_Perspective_of_Intellectual_Property_for_AI- Created_Works accessed 27 April 2026.

6 Generative -AI and Copyright Law Practices: Indian Perspective

7 Copyright Act 1957 (India).

8 Generative -AI and Copyright Law Practices: Indian Perspective

9 Sahibpreet Singh and Manjit Singh, ‘Artificial Intelligence and Intellectual Property Rights: Comparative Transnational Policy Analysis’ (2026) https://www.researchgate.net/publication/394541109_ARTIFICIAL_INTELLIGENCE_AND_INTELLECTUA L_PROPERTY_RIGHTS accessed 27 April 2026.

10 NITI Aayog, National Strategy for Artificial Intelligence (Government of India, 2023)

https://www.niti.gov.in/sites/default/files/2023-03/National-Strategy-for-Artificial-Intelligence.pdf accessed 27 April 2026.

Ø Doctrinal inadequacy of existing IP laws; Ø Divergence in proposed ownership models; and Ø Emerging consensus on the need for reform.

THE RISE OF AUTONOMOUS AI AND LEGAL DISRUPTION

Autonomous Artificial Intelligence (AI) is doing more than ever before, and often doing it cheaper, faster, and better than humans are11. Unlike traditional AI, modern generative AI does not merely assist human decision-making but also produces content with minimal human intervention. This transition from AI as a tool to AI as a creator fundamentally challenges the assumptions underlying intellectual property law12. As said, in today modern AI systems there is minimal human intervention, they are trained on vast datasets, which enables them to identify patterns and give outputs based on probabilistic predictions. This has been reiterated in his thesis by Ryan Abbott13 too. Similarly, the World Intellectual Property Organization14, says that AI is a new digital frontier that will have a profound impact on the world, transforming the way we live and work.

It is particularly reshaping the creativity and invention required in industries reliant on data-intensive processes15. Stephen Thaler invented the AI machine known as “Device for Autonomous Bootstrapping of United Sentience (DABUS)”, to generate inventions16. It raised a critical question of whether an AI can be recognised as an inventor to apply for patent. DABUS has claimed to be an inventor in patent applications across more than 15 countries17. The DABUS patent dispute has brought to light a significant gap in India’s patent regulations regarding AI-generated ideas18. 11 Ryan Abbott, The Reasonable Robot: Artificial Intelligence and the Law (Cambridge University Press, 2020). 12 Ryan Abbott, The Reasonable Robot: Artificial Intelligence and the Law (Cambridge University Press, 2020).

13 Ryan Abbott, The Reasonable Robot: Artificial Intelligence and the Law (Cambridge University Press 2020). 14 World Intellectual Property Organization, WIPO Technology Trends 2019: Artificial Intelligence (WIPO

2019) 15–20.

15 World Intellectual Property Organization, WIPO Technology Trends 2019: Artificial Intelligence (WIPO

2019) 15–20.

16 ‘The Latest News on the DABUS Patent Case’ (IP STARS) https://www.ipstars.com/NewsAndAnalysis/thelatest-news-on-the-dabus-patent-case/Index/7366 accessed 27 April 2026.

17 Khurana and Khurana, ‘DABUS Case: AI Inventorship in Indian Legal Regime’ (19 March 2025)

https://www.khuranaandkhurana.com/2025/03/19/dabus-case-ai-inventorship-in-indian-legal-regime accessed

27 April 2026.

18 Khurana and Khurana, ‘DABUS Case: AI Inventorship in Indian Legal Regime’ (19 March 2025)

The implication of this decision is further beyond patent law; it highlights a broader doctrinal limitation: IP law is structurally designed to recognize human creators. In the case of, Naruto v Slater19, the limitation was reinforced and the court denied copyright protection to a photograph taken by a monkey on the ground that non-human entities cannot be authors20, and in the case of Acohs Pty Ltd. v Ucorp Pty Ltd. 21, the court held that machine-generated works lacking human authorship are not entitled to copyright protection22. Enter 2021: India grants for the first time the recognition of Suryast, which is an artwork produced by an AI model called RAGHAV (Robust Artificially Intelligent Graphics and Art Visualizer) and Ankit Sahni23.

While the move was revolutionary and progressive, as it came right after South Africa became the first country to issue a patent designating an AI tool- DABUS-as the inventor, in 2023 the IP office withdrew it24. The same raises various jurisdictional concerns as well as other legal concerns. Therefore, the more the rise in autonomous AI, the more is the structural tension within IP law: where the law is trying to protect and reward creativity and innovation, but it no longer reflects technological reality. This shows that it is necessary to review, return and re-examine the doctrinal and foundational principles, particularly in jurisdictions like India where there is a gap in the statutory frameworks which have not yet adapted the realities of AI-driven creation.

OWNERSHIP MODELS FOR AI-GENERATED WORKS

Human-Centric Ownership Model Human-centric model is not only a theory of authorship, but at an advanced level it is best understood as a theory of risk-allocation and governance mechanism. It enables courts and regulators to anchor liability, enforcement, and licensing in identifiable legal https://www.khuranaandkhurana.com/2025/03/19/dabus-case-ai-inventorship-in-indian-legal-regime accessed

27 April 2026.

19 Naruto v Slater 888 F.3d 418, 426–27 (9th Cir 2018).

20 Naruto v Slater 888 F.3d 418, 426–27 (9th Cir 2018).

21 Acohs Pty Ltd v Ucorp Pty Ltd [2012] FCA 42, [63]-[65].

22 Acohs Pty Ltd v Ucorp Pty Ltd [2012] FCA 42, [63]-[65].

23 ‘From Original Photo to AI Art: Suryast Lacks the Required Human Touch for Copyright Protection’ (Baker Botts) https://ourtake.bakerbotts.com/post/102ivxv/from-original-photo-to-ai-art-suryast-lacks-the-requiredhuman-touch-for-copyrig accessed 27 April 2026.

24 Ed Conlon, ‘DABUS: South Africa Issues First-Ever Patent with AI Inventor’ (Managing IP, 2021)

https://www.managingip.com/article/2a5bqo2drurt0bxl7ab1q/dabus-south-africa-issues-first-ever-patent-withai-inventor accessed 27 April 2026.

persons. It operates by giving them ex-post ownership. As Ryan Abbott25, argues that the human-centric model collapses the distinction between creative labour and capital investment, thereby subtly transforming IP from a reward into a tool of technological control. This model states that ownership is only meant for those parties who prompts rather than substantive intellectual contribution. In the Indian context, where access and competition are major policy concerns, such effects might have significant distributive consequences. • AI-as-Author or Inventor Model The major challenge with AI-as-author model is its incompatibility with the normative architecture of IP law and is often dismissed on formal grounds. IP law is a regime structured around intentional agency, responsibility, and incentive alignment.

If we recognise AI as an author, we will have to reconceptualize a lot of doctrinal and foundational principles, especially the link between rights and accountability. In the case of Thaler v Vidal26, the court held that inventorship carries legal consequences that needs someone who is capable of being tried, and on whom liability can be laid down. Extending authorship to AI would therefore require a fundamental reconfiguration of rights, including ownership, transferability, and enforcement- an outcome most scholars like Jane C. Ginsburg27 view as impractical. She further contended that authorship embodies a moral and expressive dimension that cannot be replicated by algorithmic systems28. The significant challenges to this model are: Ø AI lacks legal personality; Ø It cannot hold rights or obligations; Ø It cannot be held accountable.

25 Ryan Abbott, The Reasonable Robot: Artificial Intelligence and the Law (Cambridge University Press 2020).

26 Thaler v Vidal 43 F.4th 1207, 1211–12 (Fed Cir 2022).

27 Jane C Ginsburg, ‘Jane C Ginsburg’ (Columbia Law School) https://www.law.columbia.edu/faculty/jane-cginsburg accessed 27 April 2026. 28 Jane C Ginsburg, ‘Jane C Ginsburg’ (Columbia Law School) https://www.law.columbia.edu/faculty/jane-cginsburg accessed 27 April 2026.

Public Domain Approach The implications of the public domain model are more complex in practice. It suggests that AI-generated works should not receive IP protection and should be freely accessible. It avoids forcing non-human outputs into human-centric frameworks. By denying protection altogether, this model may lead to underproduction of high-cost AIgenerated outputs, especially in sectors like pharma, design, and media, where huge amount of investment is required. As stated by Ryan Abbott29, absence of exclusive rights could weaken incentives for both AI development and deployment. All of this creates a tension between doctrinal clarity and economic efficiency. • Sui Generis Regime The main focus of the sui generis model is to design a function-specific regulatory framework which is post-doctrinal in approach.

It seeks to balance innovation and access instead of focusing only on authorship. Many policy discussions within the World Intellectual Property Organization30, suggest that the sui regime model frameworks could include limited-duration rights, mandatory disclosure obligations, and calibrated exceptions to prevent over-concentration of ownership31. It also allows multiple jurisdictions like India to address AI-generated works without any disruption to the current conceptual foundations of IP law. Emergence of data-driven monopolies can be prevented while incentivizing innovation are the safeguards like compulsory licensing, competition oversight, and public interest exceptions. Across all the four models discussed above, the main issue is the reconfiguration of IP law from a system that governs machine-generated value along with reward for human creativity.

The choice between models therefore reflects deeper normative commitments- between authorship and control, incentives and access, and doctrinal coherence and technological adaptation. 29 Ryan Abbott, The Reasonable Robot: Artificial Intelligence and the Law (Cambridge University Press, 2020). 30 World Intellectual Property Organization, ‘WIPO’ https://www. wipo. int/portal/en/index. html accessed 27 April 2026. 31 World Intellectual Property Organization, ‘Revised Issues Paper on Intellectual Property Policy and Artificial Intelligence’ (WIPO, 2020) https://www. wipo. int/meetings/en/doc_details. jsp? doc_id=499504 accessed 27 April 2026.

JUDICIAL INTERPRETATION

Indian courts in comparison to the international courts such as DABUS litigation have not yet directly adjudicated disputes involving fully autonomous AI-generated works. A major shift in the Indian copyright law happened in the case of Eastern Book Company v

D. B. Modak32, where the court rejected the traditional “sweat of the brow” doctrine and adopted a “modium of creativity” standard- where it stated that skill, judgment, and intellectual effort are the most important components. But, when we see these components with respect to AIgenerated works, the threshold articulated by the court fails. The emphasis on human creativity was further reiterated in the case of R. G. Anand v Deluxe Films33, where the court said that “expression of human ideas shaped through skill and judgement”. In the context of AI, outputs are emerged from algorithm processes rather than conscious human effort. Similarly, in the case of Najma Heptulla v Orient Longman Ltd. 34, the court stressed on the necessity of “substantial intellectual contribution”. The jurisprudence on moral rights further was highlighted the human-centric nature of IP law.

In Amar Nath Sehgal v Union of India35, the court said that artistic works are the extensions of the author’s personality, highlighting dignity, reputation, and personal connection. This specifically personality-based approach is incompatible with the AI-generated works, which lacks identity. Moral rights cannot be extended to such works as it will lead to disruption of the foundation of copyright law. There is a particular case which is relevant in the context of AI training on copyrighted datasets- Civic Chandran v Ammini Amma36, this case suggests that Indian law may accommodate technological development through broader exceptions. However, this flexibility is for the input stage and not the output stage of the AI-generated work.

Indian jurisprudence does not merely exclude AI-generated works by omission; it actively constructs a framework in which human authorship is a necessary precondition for legal

32 Eastern Book Company v DB Modak (2008) 1 SCC 1, [36]-[40].

33 RG Anand v Deluxe Films (1978) 4 SCC 118, [46]-[48].

34 Najma Heptulla v Orient Longman Ltd AIR 1989 Del 63, [13]-[15].

35 Amar Nath Sehgal v Union of India 2005 (30) PTC 253 (Del), [55]-[58].

36 Civic Chandran v Ammini Amma 1996 PTC (16) 329 (Ker), [15]-[17].

protection, thereby rendering autonomous AI outputs doctrinally invisible.

KEY CHALLENGES AND POLICY CONCERNS

1. Originality and Creativity: AI-generated outputs may satisfy formal originality but

lacks human creativity. This challenges the major foundation of copyright law-which also ends up creating a false positive problem. Protection in such cases risks diluting the originality threshold and transforming copyright into a system which instead of rewarding human intellectual creation it ends up protecting computational output.

2. Accountability and Liability: If AI-generated work infringe rights to anyone in the

country, determining liability of that works becomes extremely complex. Potential candidates for it are developers, users, and AI operators- but that also creates a liability gap, as assigning responsibility and claiming damages for the wrong done from developers or users may not accurately reflect their level of control over outputs. Without clarity on this aspect, enforcement of IP rights becomes uncertain and inconsistent.

3. Economic Incentives: The main incentive of IP law is rewarding creativity and

innovation. There is a balanced situation here, on one hand denying protection to AIgenerated works may reduce investment, while on the other hand granting protection may create monopolies over machine-generated content. It may also lead to underproduction of valuable AI-generated outputs, especially where private investment depends on exclusivity.

4. Moral Rights: For example: the right of attribution, is inherently human-centric.

Extending them to AI is very problematic. It cannot be meaningfully translated into a machine context, reiterating the limits of adapting existing doctrines of AI.

CONCLUSION

The challenge created by Artificial Intelligence (AI) to Intellectual Property law (IP) is not only a question of classification, but one of conceptual disruption. As discussed in this paper, the architecture of IP law- specifically in India- is embedded in human authorship, intentionality, and accountability. Autonomous and generative AI unsettles each of the above assumptions simultaneously. The result is not just a gap in the policy or statutory coverage, but

a structural mismatch between doctrine and technological reality. What comes to my mind constantly is that changing existing frameworks without re-examining their conceptual and foundational risks might distort the entire purpose of IP law itself. A purely human-centric work when compared to a output generated with no human input, increasingly operates as a legal fiction. On the other hand, recognizing AI-as-a-author or inventor would require an overall set up related to legal personality, liability, and rights allocation- an outcome which is not stable in the upcoming term. Similarly, putting such work in the public-domain may cost us the weakening of innovation incentives in capital-intensive sectors. The real issue is not which ownership model to choose, but acknowledging that AIgenerated work does not fit the justification of IP as a reward for human creativity.

In this regard, a carefully designed sui generis framework appears to offer the most viable path- which balances technological advancement with concerns of equity, competition, and public interestsparticularly in India. Indian law currently responds to AI not through the process of adaptation, but through exclusion. This position is unlikely to remain tenable with the increasing growth of autonomous AI as an integral to creation and innovation processes. Conclusively, the question is how IP law should evolve to remain relevant in a rapidly changing technological landscape. Any future law, policy or framework must have a balance between incentivizing innovation, ensuring accountability, and preserving the conceptual foundations of IP law.

For India, this moment shows an opportunity to move beyond reactive legal adjustments and towards a proactive framework- capable of governing human creativity along with machine-generated knowledge.