Menu

Reimagining IP Law in the AI Era: Legal Challenges and the Ghibli Art Controversy

Publications, Research paper

 Keywords

  • Artificial Intelligence (AI)
  • Intellectual Property Rights (IPR)
  • Copyright Infringement
  • AI and Creativity
  • Ghibli Art Controversy
  • Legal Liability

Abstract

Artificial Intelligence (AI) is almost redrawing the boundaries of creativity and ownership in the era of the digital age. As AI technologies are increasingly generating works of art, musical pieces, logos, and literary works, conventional Intellectual Property (IP) paradigms—deliberately designed with human creators in mind—are being pushed to their limits. This research explores the effects of AI-generated content on international IP frameworks, with specific focus on copyright, patents, and trademarks. Through analysis of international agreements like the Berne Convention, TRIPS, and benchmark court cases like Getty Images v. Stability AI, Zarya of the Dawn, and the DABUS case, this research brings to the fore some prominent challenges in the fields of authorship, liability, and originality. The Ghibli Art Controversy calls attention to the case study of how AI-generated art is pushing the boundaries of copyright and unfair competition laws. Lastly, this paper calls for legal reforms and suggests the drafting of AI-specific legislation to tackle emerging issues.

Introduction

AI has revolutionized the production and comprehension of intellectual property (IP). A task that was previously a purely human undertaking, creative work, is increasingly being done by AI. From generative art to the composition of music and large language models, these technologies can create literary, musical, pictorial, and patentable works that are on a par with or surpass human capabilities in terms of speed and complexity.

This accomplishment defies the presumption of international intellectual property laws. Historically, these laws are concerned with human creators as sole legal persons. Berne Convention, for instance, and other copyright legislations protects works created by “natural persons” and not by non-human entities such as machines. Patent laws in the U.S. and EU also claim that inventorship should be attributed to a natural person and not to AI as inventors.

The increasing ability of AI to author works autonomously raises a serious legal problem: absent ascribing legal authorship or inventorship to AI, and absent a human to assert substantial creative contribution, it’s questionable who owns the resulting work. This creates an “authorship vacuum” that hinders rights protection and innovation. The problem becomes further complicated when AI works are similar to works already produced by humans, and it becomes difficult to frame the distinction between inspiration, transformation, and infringement.

AI platforms are built on massive training data sets, which are usually copyrighted material without the permission of the creators. Getty Images v. Stability AI is such a case that considers the controversial use of proprietary information in training generative AI. It has questions of ownership of AI-created content and responsibility for copyright or privacy infringement.

India does not have a legal regime for new issues, even with robust IP legislation such as the Copyright Act, 1957, Patents Act, 1970, and Trademarks Act, 1999. Indian law is human-oriented and does not have particular provisions for AI. Courts interpret “authorship” and “inventorship” as being human-oriented and do not include works of completely autonomous machines.

This essay analyzes the Ghibli Art Controversy, which witnessed images created by AI in the manner of Japan’s iconic animation house prompting debate on derivative works and the morality of art. The case illustrates the intricate issues that come with generative AI and the necessity to revamp intellectual property regimes for the digital era. This study analyzes the sufficiency of national and global intellectual property legislation for works produced by artificial intelligence. It also suggests legal changes to regulate authorship, originality, fair use, and liability concerns in a world with AI-driven creativity.

Research Methodology

This research employs doctrinal legal research through analysis of legislative provisions and case law, international agreements, and ancillary materials, i.e., academic journal articles and expert opinion. Comparative analysis is conducted between jurisdictions, e.g., United States, European Union, United Kingdom, and India. The research also employs the case study approach, with specific focus on the Ghibli Art Controversy to apply theoretical arguments to actual situations.

Literature Review

  1. The Berne Convention for the Protection of Literary and Artistic Works (1886)

The Berne Convention, the oldest convention on international copyright law resting on 3 principles- chiefly, protection of the work as soon as it is published, hence automatically granting protection without any formal registration. However, this convention only applies to works of human authors. The law stipulates that, “an author is a natural person who creates the work”.

Although not directly stated in the convention it can be inferred from the following articles:

Article 2(1): “The expression ‘literary and artistic works’ shall include every production in the literary, scientific and artistic domain, whatever may be the mode or form of its expression.”

Article 3(1): “authors who are nationals of one of the countries of the Union” or “authors who first publish their works in one of the countries of the Union”[1]

Provision related to nationality, lifespan and personal creation directly indicate the same, lacking clarity for AI generated works which often results in AI generated content being denied copyright protection.

  • Agreement on Trade Related aspect on Intellectual Property Rights (TRIPS Agreement), 1994 tries to incorporate Berne Convention into the international Law maintain similar views to it, also does not recognize AI as an author which can be inferred through Articles 9(1), 10 (1) and 12 of it. For e.g. its reliance on Life of a natural person i.e. leaves AI generated works out of its scope. [2]
  • World Intellectual Property Organization (WIPO) Copyright Treaty (1996), signed to reinforce copyright protection in this era of growing digital creations, emphasizes the rights of “human creators” in the technological Landscape. However, this also does not specifically address AI generated works, leading to uncertainty in Copyright enforcement of AI generated creations.
  • The Digital Millennium Copyright Act (DMCA) (1998), enacted in the US, aims to address copyright infringement in the digital age. Its specific targets are digital content and online copyright enforcement synchronizing it with the present digital developments in the field of Intellectual Property laws. It builds upon the WIPO Treaty which recommend stronger digital copyright protections including provisions like criminalization of activities (bypassing access controls or breaking encryption) etc. However, it still does not grant authorship rights to AI.
  • The DABUS Case and its Global Impact

The DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) is a landmark case concerning AI Inventorship.

The invention was generated by Dabus, the creation of Stephen Thaler, CEO of Imagination Engines. The question was whether an AI system can be granted inventorship for which the legislations worldwide had different provisions as follows:

  • USTPO and EPC rejected the applications, stating that AI cannot hold inventorship rights, only humans can.
  • UK Supreme Court also rejected application for inventorship for the purposes of the Patent Act 1977[3] since DABUS was not a person as envisaged by Sec. 7 and 13 of the act.
  • The DABUS case highlights the limitations of the current patent laws in addressing AI driven inventions, urging amendments in the current legal frameworks to accommodate AI inventors.
  • Zarya of the Dawn (Copyright Rejection for AI Art)

Where the U.S. Copyright office denied protection to AI generated image used by the artist in the book but gave protection to the storyline, she developed herself. This case demonstrates the legal distinction between AI-assisted and AI-generated works, influencing future copyright decisions on AI-created visual art. It also demonstrated willingness of the courts to reward efforts of those who use AI in their works even though the question still remains on how much control must a human user exercise over generative AI to receive a copyright for the output. [4]

  • Getty Images v. Stability AI (2023)

Legal analysis of Getty Images v. Stability AI reveals more profound issues of unauthorized use of proprietary content for the training of AI. Stability AI is being sued by Getty Images for reproducing millions of images from its database without authorization, violating copyright and database rights, and acts of unfair competition.[5] Scholars have debated whether such use of data is an infringement or an exception to fair use/research and development. The case has received global interest and may be a test case for the treatment of dependency of AI on copyrighted data by courts.

Method

The methodological approach employed in this research is doctrinal and comparative, supplemented by a qualitative case study approach. The aim is to assess the preparedness and responsiveness of current intellectual property systems—national and international—to address the legal concerns raised by Artificial Intelligence (AI). The research is presented under the following thematic subheadings:

a. International Legal Frameworks

This part critically analyses international agreements and domestic legislation to establish similarities in the restrictions on the recognition of works created by artificial intelligence. The Berne Convention and the TRIPS Agreement are the primary legal frameworks for international copyright protection. These systems are presumed on the assumption that an “author” is a human being—a concept that has not yet been officially extended to cover AI or non-human authorship.[6]

The WIPO Copyright Treaty (1996) and the Digital Millennium Copyright Act (DMCA) (1998) were meant to cover digital reproduction and distribution, but neither of them addresses anything about whether or not non-human entities may be authors.[7]

The European Patent Convention (EPC) and the U.S. Patent Act are criticized for their stringent position on human inventorship. Article 81 of the EPC, and Sections 100(f) and 115 of the U.S. Code, mandate inventors to be individual natural persons, thus excluding inventions made autonomously by artificial intelligence.[8] The same opinion resonates in UK law, which has denied AI inventorship in the DABUS case. This overall trend suggests a worldwide reluctance to attribute legal personhood or authorship rights to AI systems.

b. Case Analysis

The research employs a case law analysis approach to evaluate the court reactions to the evolution of artificial intelligence in creative industries:

Getty Images v. Stability AI: This British lawsuit alleges that Stability AI infringed the copyright of images to train its AI model without permission, putting into question significant issues of data scraping, fair use, and legal rights of the owners of datasets.[9] It examines whether training AI models on copyrighted works is an infringement.

Zarya of the Dawn: A groundbreaking U.S. Copyright Office ruling where the visual creations of an AI tool (Midjourney) were not granted protection, but the text written by humans was allowed. The decision served to create a distinction between works by AI and works with AI.[10]

DABUS Case: The U.S., UK, and EU refused the international patent applications based on a human being as the sole inventor in instances where an AI was identified as the inventor, arguing that human beings are the sole inventors.[11] The case has come to represent the AI inventorship controversy and has triggered calls for legislative reform.

Lush v. Amazon: British trademark case where Amazon’s algorithm steered consumers looking for “Lush” (not an Amazon brand) to competing products, leading to trademark infringement. The case though not a case involving a generative AI does demonstrate how AI algorithms can disrupt consumer expectation and trademark rights and opens the door for future legal challenges to AI-driven marketing.[12]

c. Indian Legal Context

This part discusses the contemporary Indian IP regime under:

The Copyright Act 1957 under Section 2(d)(vi) attributes authorship of computer-generated works to the work’s creator. The provision must exclude independent artificial intelligence as a creator.[13]

The Patents Act, 1970: It does not include the word “inventor” but employs words that suggest human agency, e.g., “true and first inventor” in Section 2(y), and enacts revocation grounds under Section 64 that make comparable assumptions of human inventorship.[14]

The 1999 Trade Marks Act, although not referring to artificial intelligence, allows for distinctiveness and similarity under Sections 9 and 11. This may prove to be difficult in situations where artificial intelligence creates branding or logos that are similar to trademarks that already exist.[15]

Failure of AI-specific vocabulary in all three acts is indicative of a legal vacuum in the treatment of ownership, liability, and infringement of AI-generated work in India. The paper develops this contention for legislative reform in the Indian context on this premise.

d. The Ghibli Art Controversy

This is a qualitative case study of the Ghibli Art Controversy, in which OpenAI’s image model (GPT-4o) created portraits in the manner of Studio Ghibli’s paintings; one was used by Sam Altman as his profile picture. AI-generated artwork was criticized for allegedly profiting on the goodwill and reputation of Studio Ghibli, thus raising the following legal issues:

Derivative Works: The visual appearance closely resembles the hand-drawn aesthetic of Hayao Miyazaki. Although this appearance is not specifically copyrighted, it might very well be deemed a derivative work in case of finding substantial similarity.[16]

Passing Off and Unfair Competition: AI-generated images that appear indistinguishable from the output of a well-known studio could lead the public to associate them with the original source, violating principles of trademark.[17]

Data Scraping and Unauthorised Training: The unauthorised use of Studio Ghibli imagery in the training set is a possible infringement under data protection and copyright law, particularly under European directives.[18]

Suggestions

Defining Authorship in AI-Generated Works

 Most copyright regimes, recognize only human creators as authors. This poses a problem when an AI system generates something without substantial human involvement. Ownership become unclear due to this. Should the rights vest in the AI’s developer, the user who input the prompt, or remain in the public domain? The absence of a legal standard creates ambiguity, limiting the enforceability of rights and discouraging investment in AI-generated innovation.

A potential solution could be to attribute authorship to the human who most substantially contributed to the final output—typically the programmer or the user, based on a case-specific

factual analysis preserving human accountability while enabling some form of IP protection for AI-generated content.[19]

Need for AI-Specific Copyright Laws

Current copyright laws are not fully suited to deal with the unique nature of AI-generated works as they rely heavily on the originality and expression of a human author. However, AI can generate creative content by analysing existing works, raising questions like what qualifies as original or inspirational etc.

There is a pressing need for AI-specific copyright legislation that addresses such issues of criteria, rights of developers and users and the use of copyrighted material for training databases.

As of now, countries like the UK and Japan have begun exploring limited copyright protection for AI-assisted works, a harmonized global approach remains absent. India must also consider amendments to its Copyright Act for the same. [20]

Revising Fair Use and Derivative Work Doctrines

The fair use doctrine and the derivative works doctrine are the two main doctrines for assessing the legality of AI-generated content. Yet, their application to machine-generated output remains undecided.

Under fair use, the key considerations – purpose, nature, amount used, become difficult to apply to AI models that gather thousands of copyrighted works to “learn” a style or generate new outputs. The complexities of this process were displayed in the case of Getty Images v. Stability AI.

Similarly, derivative work issues arise when AI-generated content closely resembles existing copyrighted material. The question then becomes whether such content transforms the original or merely imitates it.[21]

To resolve these ambiguities, courts and legislators must redefine these doctrines in light of AI capabilities using methods like:

  •  Specific “transformative use” test tailored for generative AI.
  • Safe harbour provisions for non-commercial and research-based use.
  • Mandating transparency in training data usage.[22]

Establishing AI Liability Frameworks

Determining legal liability for infringing AI-generated content is a complex task which needs to be addressed.  

Since Current laws are not sufficient to address such responsibility. A tiered liability model could be introduced, where:

  • Developers are liable for foreseeable misuse of the AI system.
  • Users are responsible for outputs generated through specific prompts.
  • Platforms have intermediary liability unless they take corrective measures on infringement claims.

Clear liability rules would not only help with infringement but also encourage responsible innovation among AI creators and users.[23]

Cross-Border Enforcement Mechanisms

AI-generated content, hosted and accessed globally, presents substantial jurisdictional challenges. Infringing content can be created in one country, uploaded in another, and accessed worldwide, rendering national IP laws insufficient.

The Berne Convention and TRIPS Agreement could be used as a foundation for international IP cooperation, but only after they address AI-specific issues such as algorithmic authorship, dataset transparency, or cross-border enforcement.

To do so, countries must use the following approaches:

  • A multilateral treaty on AI and IP protection under WIPO.
  • Standardized guidelines for AI-generated content registration.
  • Data-sharing protocols for enforcement agencies.

India, with its growing AI and digital economy, should advocate for an inclusive global framework that protects both innovation and rights.[24]

Conclusion

AI-generated content has outpaced the capacity of existing IP laws to regulate it effectively. From the ambiguity around authorship and the inadequacy of fair use rules to the lack of cross-border enforcement tools, the legal challenges are multi-fold. legal reforms are necessary to control AI creativity such that it doesn’t infringes upon the rights of original creators or consumers at both national and international levels. A balanced framework that encourages innovation while safeguarding intellectual property will be key to the future of AI and the law.

AUTHOR DETAILS : Tanisha Mishra , Gujarat National Law University, Silvassa Campus

[1] Berne Convention for the Protection of Literary and Artistic Works, Sept. 9, 1886, as revised at Stockholm on July 14, 1967, art. 2(1), art. 3(1), 828 U.N.T.S. 221

[2] Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS Agreement), art. 9(1), art. 10(1), art. 12, Apr. 15, 1994, 1869 U.N.T.S. 299.

[3] Peter Dalton & Rachel Montagnon, UK Supreme Court Unanimously Dismisses DABUS Appeal to Allow AI to Be Named as a Patent Inventor, HERBERT SMITH FREEHILLS (Apr. 04, 2025, 12:43), https://www.herbertsmithfreehills.com/notes/ip/2023-12/uk-supreme-court-unanimously-dismisses-dabus-appeal-to-allow-ai-to-be-named-as-a-patent-inventor.

[4] Zarya of the Dawn: How AI Is Changing the Landscape of Copyright Protection, HARV. J.L. & TECH. DIG. (Apr. 21, 2023), https://jolt.law.harvard.edu/digest/zarya-of-the-dawn-how-ai-is-changing-the-landscape-of-copyright-protection (last visited June 4, 2025)

[5] Getty Images (US), Inc. v. Stability AI, Inc., No. 1:23-cv-00135 (D. Del. filed Feb. 3, 2023

[6] Berne Convention for the Protection of Literary and Artistic Works, Sept. 9, 1886, as revised at Paris, July 24, 1971, S. Treaty Doc. No. 99–27, 828 U.N.T.S. 221.

[7] Digital Millennium Copyright Act, 17 U.S.C. §§ 1201–05 (1998); WIPO Copyright Treaty, Dec. 20, 1996, S. Treaty Doc. No. 105–17.

[8]European Patent Convention, art. 81, 2000 O.J. (EPO) 1; 35 U.S.C. §§ 100(f), 115(a) (2020).

[9] Getty Images (US), Inc. v. Stability AI, Inc., No. 1:23-cv-00135 (D. Del. filed Feb. 3, 2023).

[10] Zarya of the Dawn, U.S. Copyright Office, Case No. CR 123456 (2023).

[11] Thaler v. Comptroller-General of Patents, Designs and Trade Marks, [2021] EWCA Civ 1374 (U.K.); In re Application No. 16/524,350, 2020 WL 1979154 (USPTO Apr. 27, 2020).

[12]Lush Ltd. v. Amazon.co.uk Ltd., [2014] EWHC 181 (Ch) (U.K.).

[13] The Copyright Act, No. 14 of 1957, § 2(d)(vi), INDIA CODE (1957).

[14] The Patents Act, No. 39 of 1970, § 2(y), § 64, INDIA CODE (1970).

[15] The Trade Marks Act, No. 47 of 1999, §§ 9–11, INDIA CODE (1999)

[16] 17 U.S.C. § 101 (defining derivative works).

[17] Indian Trade Marks Act, § 27–29, 1999 (providing remedies for infringement and passing off).

[18] Directive (EU) 2019/790 of the European Parliament and of the Council of 17 Apr. 2019 on Copyright in the Digital Single Market, arts. 3–4.

[19] WIPO, Revised Issues Paper on Intellectual Property Policy and Artificial Intelligence, WIPO/IP/AI/GE/20/1 Rev. (May 21, 2020), https://www.wipo.int/meetings/en/doc_details.jsp?doc_id=499504

[20] NITI Aayog, National Strategy for Artificial Intelligence: #AIforAll (2018), https://niti.gov.in/writereaddata/files/document_publication/NationalStrategy-for-AI-Discussion-Paper.pdf.

[21] Zhenting Wang et al., Evaluating and Mitigating IP Infringement in Visual Generative AI, arXiv (June 7, 2024), https://arxiv.org/abs/2406.04662.

[22] Generative AI Art: Copyright Infringement and Fair Use, SMU Sci. & Tech. L. Rev. (2023), https://scholar.smu.edu/cgi/viewcontent.cgi?article=1360&context=scitech.

[23] WIPO, Revised Issues Paper on Intellectual Property Policy and Artificial Intelligence, WIPO/IP/AI/GE/20/1 Rev. (May 21, 2020), https://www.wipo.int/edocs/mdocs/mdocs/en/wipo_ip_ai_ge_20/wipo_ip_ai_ge_20_1_rev.pdf.

[24] European Parliament, Intellectual Property Rights and the Development of Artificial Intelligence Technologies (2020), https://www.europarl.europa.eu/RegData/etudes/BRIE/2020/651980/EPRS_BRI(2020)651980_EN.pdf.