INTELLECTUAL PROPERTY AND ARTIFICIAL INTELLIGENCE: THE DICHOTOMY OF OWNERSHIP

-       Kriti Priyadarshini^[1] & Mrinal Litoria^[2]

                                                                                                                                             ABSTRACT

The rapid integration of artificial intelligence into creative and inventive processes seemingly poses a profound challenge to the foundation of modern intellectual property law, necessitating evaluation of legal rights and responsibility. The creative capacity of machines seems to challenge the very bedrock of intellectual property law which vest IP rights in the natural or legal person who creates the intellectual property subject to assignment, licensing or other contractual obligation. This article attempts to clarify that there is no real dichotomy of ownership while understanding that AI in the modern world is best treated as a sophisticated tool like any other machine used in the production of goods and/or services. An AI tool’s capacity for production does not diminish a human’s indispensable role in conception, validation and commercialization of any produced IP. A sincere effort has been made to examine the global jurisprudence and discuss major outlooks in light of the legal history of intermediated creation. The resulting policy imperative is clear: to ensure the responsible use of AI and preserve the integrity of the IP system, full legal and commercial liability for infringement, defective disclosure, and other IP wrongs should, as a matter of policy, be assigned solely to the human user, and likewise, any IP rights should solely vest in the human user.   

Keywords: Artificial Intelligence, Copyright Law, Intellectual Property Policy.

                                                                                                                                              INTRODUCTION

For centuries creation and protection of intellectual property have been defined by the relationship between the “guiding will” and the “executing hand”. Historically, assistance for creation of protectable IP has been sought from commissioned artists, research assistants and ghost writers who have provided the executing labour, while the legal rights and responsibilities have accrued to the individual or legal body who furnished the instructions and upon whose instructions the work was ultimately executed. The rise of AI is representative of the most advanced iteration of this dynamic, functioning as a most sophisticated intermediary whose speed and complexity substantially reduces manual human labour.

This article argues that AI’s role in the creation of any IP is that of an intelligent intermediary. We are at such a stage of technological advancement which now presents before us a crucial opportunity for the legal system to formalize a necessary framework for human stewardship and robust human centred liability. IP law is at its core a very human centric law, and this article seeks to preserve the core nature of IP law by establishing that the human user’s decisive interventions – from initial prompt to final commercial deployment are legally sufficient for “ownership”. Simultaneously, we argue that this ownership of IP should be inextricably linked to enhanced accountability by assigning legal risk to the human user, which this article argues is the most viable path to cultivate responsible IP use and safeguard the integrity of the innovation ecosystem.

THE NORMATIVE FRAMEWORK

The foundational principles of international IP law remain steadfastly rooted in rewarding human intellect. Treaties like the Berne Convention^[3] which emphasizes moral rights and the requirement laid down in the TRIPS Agreement^[4] for an “inventive step” have been interpreted as presupposing the presence of a human mind capable of making moral choices and conceptual leaps. At present, the job performed by AI is that of computation, it does not possess any “intent” or “judgment” – which are elements foundational to all IP rights.

Much of the current legal debate risks exaggerating AI’s autonomy. The reality is the AI’s output is not final; in fact all AI data requires human validation for it to be used in the context of creation of any protectable IP. True legal leverage resides in the human capacity to define the problem, evaluate the suggestions and solutions provided by AI and most importantly provide the ethical and commercial context that an AI system lacks. The normative framework thus rests on a crucial yet delicate division of labour – AI for processing, humans for judgment and legal responsibility.

COPYRIGHT AND PERSONALITY RIGHTS

The debate over “authorship” in the context of copyright must pivot from questioning AI’s contribution to affirming the legal sufficiency of a human’s guiding will. Human input must not be considered secondary, in provides the defining creative direction.

a)    Originality through curation and fixation

Under many copyright regimes, standards for originality required for copyrightability for any material can be met through the human user’s decisions and instructions:

I.      Selection and instruction: The human user defines the horizons of the creative universe through a combination of prompts and parameters.

II.    Filtering and refinement: The human reviews the output, critiques and selects which parts to retain, resembling an editor choosing and refining passages written by a ghost writer.

III.  Finalization and commercialization: The raw output is finalized and transformed into a published finished work.

These steps may satisfy the minimal creative contribution required by many jurisdictions and are consistent with precedents that reward skillful selection and arrangement such as Eastern Book Co. v. D.B. Modak.^[5] For the legal system to keep up and be relevant in the current technological environment, jurisprudence should identify and recognize that the true source of protectable expression is the human’s guiding intent which is expressed through complex prompt engineering and final editing. Such an understanding affirms the human as the author who satisfies the requirements of the Indian Copyright Act, 1957.^[6] In Li v. Liu,^[7] the Beijing Internet Court held that an AI-generated image met the originality threshold because it reflected the plaintiff’s “intellectual investment” in configuring the model and selecting the final output, thus correctly recognized the indispensable intellectual investment by the human.

It is also important to acknowledge the stance taken in the USA by the District Court for the Distrct of Columbia in Thaler v. Perlmutter^[8] wherein an application was made to register copyright for an image, in which he listed a generative AI system “Creativity Machine” as the author to an image titled “A Recent Entrance to Paradise”. In the decision, it was affirmed that “human authorship is a bedrock requirement of copyright”, thereby denying protection to outputs that were solely conceived by an algorithm. This position was later affirmed by the D.C. Circuit in Thaler v. Perlmutter which held that the US Copyright Act’s concept of ‘author’ refers only to human beings, thereby denying registration to a work created by an AI system.

This position was further reflected in the U.S. Copyright Office in the Zarya of the Dawn^[9] review, which concluded that while the selection and arrangement of images along with the text written by Kashtanova were protectable however the individual images generated by Midjourney were not found to be worthy of copyright protection.

To further simplify the proposition that copyright should subsist in the human user, a simple example of creation of a computer program is relevant. Every time a new computer program is created and applied for copyright protection, the creator of computer is not enlisted as the author, neither is the creator of the language in which the code is written accorded any authorship. It is the human user who worked on developing the computer program who is the author/ owner of the program, and the tools – which is the computer and/or the coding language, are not accorded any credit in the copyright application. Likewise, in cases where AI is used as the tool to generate original content, copyright in such material must vest in the human, and not in the AI model or the owner of the AI model.

LIABILITY FOR CREATIVE WRONGS

With the rapid increase in the use of AI and the multiple use cases to which AI can be applied, it is important to ensure that all such use is done in a responsible manner. This article advocates a strong presumption of human user-side liability as the most effective policy to ensure the responsible use of AI. Placing primary responsibility on the human who issues the command and decides to publish any output that is infringing in nature, eliminates the legal incentive to use AI as a collaborator to shield oneself from liability, and also streamlines the processes for actions in copyright infringement cases.

Similarly, in cases involving the unauthorized use of persona (for eg- deepfakes), infringement is committed not by the AI, but by the human who initiates and deploys the malicious or misleading use of the AI tool. Indian courts have signalled a willingness to protect in personality rights cases protecting against unauthorized commercial use, such as Anil Kapoor v. Simply Life India^[10]. It is important to point out that the Courts have not refrained from relying on the tort of passing off and the right to publicity to protect a human subject when traditional copyright mechanisms are insufficient against non-consensual AI exploitation.

For example – every time a murder is committed using a knife from XYZ brand, XYZ brand is not held liable of such a crime along with the doer or the human responsible; it is only the human user of the knife who has committed the crime wjp is held liable. Likewise, it would be illogical to implicate the AI tool for any wrongs that re committed by the human who initiates and wrongfully uses AI to commit a civil or criminal wrong.  

                                                                                                                                              PATENTS

In patent law, the indispensable requirement for being an inventor is the “mental act of conception”^[11]. While use of AI can accelerate data analysis, current AI systems are genrally understood as being incapable of autonomously forming the definitive, non-obvious idea required for inventorship.

AI’s role in the inventive process – as demonstrated in the DABUS litigation^[12] is that of a sophisticated data processor, whereas the human’s role is to:

     I.         Define the problem and set the parameters for the AI’s search/ contribution;

   II.         Recognize the utility of the solution suggested by the AI;

 III.         Transform the AI’s output into legally valid claims that meet sufficiency requirements before the relevant office.

The true “inventive step” of recognizing the utility factor and subsequent legal framing is therefore performed by the human. AI on the other hand is simply the tool that accelerated the experimental process. Before a patent office, the human inventor should be the signatory to the patent application and must bear the full legal and financial risk for any subsequent finding of non-inventorship, insufficient disclosure or infringement. This helps prevent the opaque nature of AI from creating a liability vacuum and ensures the integrity of the patent system’s disclosure mandate upholding the quid pro quo nature of patent registration.

It is pertinent to discuss an influential but still currently a minor view that advocates for the creative autonomy of AI. Proponents of this view argue that the sophisticated and often unpredictable outputs of generative models fulfill the spirit, if not the letter, of inventorship and authorship. They state that if AI’s contribution meets the objective threshold of novelty or originality that are applied to human works, in such cases denying protection simply because the “creator” lacks consciousness is an arbitrary and economically harmful decision which actively discourages investment in fully autonomous AI systems.

These views surfaced prominently in commentary around DABUS litigation, where some examiners and commentators warned that denying patents to AI-derived inventions could push valuable technical information into trade secrecy, weakening the patent system’s disclosure function.

                                                                                                                                              TRADEMARKS

AI models train on millions of logos and brand assets, posing a material risk of outputting marks that are confusingly similar to existing ones. Such a potential flood of synthetic marks may place a substantial burden on trademark authorities and clearance searches, rapidly eroding the available pool of truly distinctive trademarks. However, ownership of trademarks is based on commercial use or its registration, not authorship; thus a human’s decision to use an AI generated trademark in commerce or registration of the same, is the focal point of liability of confusion or dilution, and not the act of creation of the trademark itself. In any case, it has always been the responsibility of the trademark owner to conduct adequate clearance and due diligence prior to adopting a trademark in trade.

PRIVATE CONTRACT LAW

The prevailing use of private contract law to govern AI use must be supported and encouraged as a necessary policy measure to insulate the AI infrastructure layer.

AI model developers (eg: Google, OpenAI) should be primarily responsible for any infringement occurring at the infrastructure layer (eg: training data infringement), while the human users who seeks direct commercial gain from the outputs created by AI, should accept the responsibility for infringement arising from their particular use of the AI generated output. Inclusion of indemnification clauses and quitclaim-style assignments^[13] correctly allocate the risk to the party best positioned to mitigate such risk, which is the human user. Such mechanisms can help ensure that the human retains meaningful financial responsibility for the IP they choose to commercialize.

                                                                                                                                              CONCLUSION

The era of AI does not necessitate a radical restructuring of IP law; it requires a focused legislative effort to fortify the foundations of existing IP law through clearer rules of human accountability. The core thesis – that for IP purposes AI should be treated as an electronic tool provides the legal justification for continuing to assign IP rights exclusively to the human user.

Maintaining the integrity of the IP system now solely depends on proactive legislative action that places clear and, where appropriate, strict responsibility on the human users in cases of wrongs committed by usage or registration of an IP materials created with the help of AI. By formalizing the human’s role as the indispensable and ultimately accountable legal entity, the law can prevent the systemic misuse of these powerful tools.

^[1] Kriti Priyadarshini is an advocate.

^[2] Mrinal Litoria is an advocate.

^[3] Berne Convention for the Protection of Literary and Artistic Works art. 6bis, Sept. 9, 1886, as revised at Paris on July 24, 1971, 828 U.N.T.S. 221.

^[4] Agreement on Trade-Related Aspects of Intellectual Property Rights art. 27, Apr. 15, 1994, Marrakesh Agreement Establishing the World Trade Organization, Annex 1C, 1869 U.N.T.S. 299

^[5] Eastern Book Co. v. D.B. Modak, (2008) 1 SCC 1 (India).

^[6] The Copyright Act, 1957, No. 14, Acts of Parliament, 1957, § 2(d) (India).

^[7] Li v. Liu, (Beijing Internet Ct. Nov. 27, 2023).

^[8] Thaler v. Perlmutter, 687 F. Supp. 3d 640, 642 (D.D.C. 2023).

^[9] Letter from U.S. Copyright Office to Van Lindberg re: Zarya of the Dawn (Feb. 21, 2023).

^[10] Anil Kapoor v. Simply Life India, CS(COMM) 652/2023 (Delhi H.C. Sept. 20, 2023).

^[11] Townsend v. Smith, 36 F.2d 292 (C.C.P.A. 1929).

^[12] Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022); Thaler v. Comptroller-General of Patents, Designs and Trade Marks 2023 UKSC 49.

^[13] See, e.g., OpenAI, Terms of Use, OpenAI (Nov. 14, 2023), https://openai.com/policies/terms-of-use (last visited Nov. 28, 2025).