Name of author: Ananya Sharma and Arjun Singh (II Year, Dr. Ram Manohar Lohiya National Law University, Lucknow)

Keywords- AI, IP Right Infringement, Ghibli, Moral Rights, Imitation

The ‘Ghibli-Style’ Conundrum: When Does Imitation Become Infringement?

Consider this: an artist presents a painting – rich landscapes, muted watercolour, and a child-like whimsy.  Praise calls it “new,” but the reference is obviously Studio Ghibli. Now, envision an AI doing this at a humongous scale. At what point does inspiration become imitation, and when does a copy begin to rob the original of its humanity?

This viral “Ghiblification”  trend, where the AI creates art mimicking Studio Ghibli’s distinctive art form, demonstrated both the creative power of AI and the practical risks of unlicensed style-replication.  The advent of generative artificial intelligence has significantly simplified the replication of the distinctive aesthetic of Studio Ghibli films, characterised by lush landscapes, evocative watercolour palettes, and iconic characters such as Totoro. This technological advancement raises a critical legal question: at what point does artistic inspiration derived from Studio Ghibli’s style cross into copyright infringement under Indian law? The resolution of this issue hinges on the distinction between permissible inspiration and unauthorised reproduction.

Central to this analysis is the idea- expression dichotomy, a principle articulated by the Supreme Court of India in R.G. Anand v. Delux Films. The Court ruled that copyright law does not cover expansive ideas, general themes, or broad concepts. For example, Studio Ghibli’s themes, including magical realism, concern for the environment, or child-like innocence, are not copyrighted and may be used without restriction. To be clear, neither the ideas presented in those examples nor those associated with those themes are copyrighted; therefore, anyone is free to appropriate other artists’ themes and ideas. However, once an idea is expressed in a particular expression, copyright protection is attached to the character design of Totoro or the bathing house of Spirited Away. Copying those specific elements would be infringing copyright because the particular expression is now copyrighted, and merely using the protected expression without permission, designed to be countercultural to the existing character of the story, would infringe copyright.

This principle was further elucidated in Eastern Book Co. v. D.B. Modak, where the court stressed that copyright protection requires a “minimum level of creativity.” Although effort and skill contribute to a work’s value, creativity drives originality. In the scenario of AI-generated art following the stylistic choices of Studio Ghibli, those generic watercolour backgrounds with soft hues may fall in the category of unprotected ideas, but if the output closely resembles the particular compositions, character representations, or visual proportions of Ghibli films, such work constitutes a “derivative work” that may infringe on the work of Ghibli’s original copyright.

The Indian Copyright Act, 1957, guides the statutory underpinning for this analysis.  Section 14 identifies the copyright owner’s exclusive rights, including the exclusive right to reproduce and adapt their works. Section 2(p) identifies derivative works as substantially derived from or using the original. 

Fair Use or Derivative Work? The AI-Copyright Conundrum in India

 All this discussion and contemplation about the growth of generative AI has led us to a pertinent yet ambiguous question: When does using copyrighted material for AI training become an act of infringement? This issue is especially pronounced in India, where the legal framework is grappling with the distinction between permissible use and illegal copying. Indian copyright law has a much narrower “fair dealing” principle than the United States’ flexible “fair use” doctrine. A work can be used without permission only for specific purposes, including private study, research, and criticism. Large-scale, unauthorised ingestion of copyrighted works for commercial purposes to train AI will have difficulty fitting into these narrow exceptions.

This raises a fundamental legal paradox: if something new was created based on an existing work, it may be considered an infringing “derivative work.” AI developers may argue that they are developing “transformative” works because AI systems learn from, rather than copy, patterns or structures. However, this is likely less of an effective line of defence in the Indian legal system.

Conversely, critics contend that AI outputs “substantially similar” to a creator’s work are unauthorised adaptations that violate the original owner’s exclusive rights. Critics have also claimed that when AI outputs mirror distinctive elements of a creator’s work, they risk being treated as unauthorised adaptations, since the underlying training process may still embed and reproduce protected expressions. The ongoing lawsuit between ANI Media and OpenAI in the Delhi High Court is a landmark case that will determine how the law applies. The outcome will set a precedent for balancing technological innovation with the creators’ rights and providing clarity in a legal grey area.

The Threat to Moral Rights: Protecting the Integrity of the Artist and Their Legacy

While talking about derivative works is relevant for indicating the tension between creativity and copyright law, another fundamental aspect of an artist’s personal interests in their work is their moral rights. Moral rights represent the individual rights of an author or creator to assert authorship of their work and ensure it is not distorted, mutilated, or modified in a manner that would be detrimental to their honour or reputation.  They help to recognise the artist’s personal connection with their work and legacy. Section 57 of the Indian Copyright Act, 1957 recognises the moral rights of an author, including the right to attribution, and ‘rights to not have their work altered (what the Act calls a distortion, mutilation or modification) in any way that is detrimental to their honour or reputation. 

Moral rights are integral and inalienable, as affirmed in Amarnath Sehgal v Union of India which provided the basis for recognising an artist’s moral rights to preserve their ability to protect their creative identity from unapproved alterations and changes in context. The introduction and use of AI-generated art is a real challenge to protect moral rights because the AI programs can reproduce a picture of an artist’s work, or even enhance it, without the artist’s consent.

Scholars argue that tools like DALL-E may generate derivatives that violate an artist’s right of attribution, and such derivatives may falsely insinuate the artist’s name. The absence of AI language in Section 57 creates confusion as we have seen with discussions about whether AI training on copyrighted works distorts the works. Recent case law, such as ANI v. OpenAI, raises concerns over AI scraping and its possible intersections with moral rights. To protect the integrity of artists, revisions to Section 57 must clarify AI’s implications, not only to require transparency surrounding training data, but also to prohibit the unconsented use of styles to exploit and misappropriate artists’ legacies as we transition into and through a new digital era.

Legislative Landscape and Future of AI: Navigating a New Frontier in Copyright Law

The challenges posed to moral rights by AI use also reveal a larger structural gap: India’s current copyright framework (The Copyright Act, 1957 ) is ill-equipped to handle generative AI technology, leading to legal uncertainties regarding authorship, ownership, and infringement. Section 2(d) describes “author” as human and does not address AI, making these works legally uncertain. The inadequacy appeared in the high-profile case of Amar Nath Sehgal v Union of India, which dealt with moral rights but did not address a situation in which the author is a non-human entity.

The provisions for fair dealing in the Act are narrow in scope, contained in Section 52 and are inadequate to facilitate the possible range of usages in training of AI data. Section 79 of the Information Technology Act, 2000 strictly limits liability for intermediaries, further complicating enforcement. The NDTV v. OpenAI case highlights the gaps. A 2025 DPIIT panel has recommended reforms to the Copyright Act that promote transparency in AI training and opt-out options for data collection, modelled on the EU model and its AI Act. The reforms contemplate agreements for data licensing that would provide value to data, which might lead to a ‘significant human input‘ test that indicates copyrightability of AI works. As a member since 1928 of the Berne Convention for the Protection of Literary and Artistic Works, India should not fall behind its international peers in adopting best practices to protect works that will undoubtedly have multiple influences and data projections. 

India risks falling behind in AI governance without reforms, mired in endless cases and delays. Finally, the proposed amendments of existing copyright laws in India should also ensure the ethical use of data, clear authorship of creators, and alignment with international standards while balancing the interests of innovators with those of creators. As AI blurs the line between homage and imitation, Indian Copyright law must advance to ensure it continues to protect the balance between creative freedom and artistic integrity. The challenge is to respect innovation, which needs to be respected. If we do not improve, India will fall behind in protecting its creators today.