The Ghibli Trend: Inspiring Creativity or Igniting Copyright Controversy?

With its launch on March 31st, OpenAI experienced an unprecedented surge in user activity. CEO Sam Altman revealed that ChatGPT gained one million users within just an hour- matching its usual five-day growth rate. As the Ghibli-style images flooded social media, they caught the attention of celebrities, influencers, and even world leaders, who eagerly joined in by sharing AI-generated portraits of themselves in the beloved animation style.

While the movement has been widely praised for its artistic appeal and nostalgic charm, it has also reignited a heated debate on copyright, intellectual property rights, and the implications of AI-generated creativity. At the core of this controversy is a pressing legal question: Does using copyrighted material to train AI models constitute copyright infringement? The answer is far from clear, as it involves complex issues of fair use, transformative works, and the evolving role of artificial intelligence in artistic creation.

The Ghibli trend primarily revolves around AI-generated images and videos that emulate the visual style pioneered by Studio Ghibli. AI models such as ChatGPT, MidJourney, Stable Diffusion, and RunwayML enable users to create highly detailed illustrations that capture the soft color palettes, painterly backgrounds, and ethereal lighting characteristic of Ghibli films. These tools rely on deep-learning algorithms trained on vast datasets, which, in many cases, may include copyrighted works. While these adaptations are largely non-commercial and serve as artistic tributes, they raise concerns about the ethics of using copyrighted visual styles and whether such works qualify as derivative or transformative creations.

Legal scholars remain divided on the issue, with conflicting opinions on whether AI-generated images qualify as fair use. The key distinction lies in the nature of AI-generated content, which differs from traditional transformative works such as parodies or critiques. Unlike human artists who intentionally reinterpret styles, AI models rely on statistical patterns to generate content, raising questions about whether they can be considered genuinely "creative." Proponents of AI-generated art primarily invoke the fair use doctrine, which permits limited use of copyrighted material under specific conditions. Fair use is generally assessed based on four key factors:

If an AI-generated image is considered "transformative," meaning it adds new meaning, context, or expression to the original work, it may qualify for fair use. For instance, Google’s Authors Guild v. Google (2015) case established that scanning books for indexing and search purposes was fair use because it transformed the way users accessed information. However, applying this logic to AI-generated art is contentious, as many argue that AI merely remixes existing data rather than creating something genuinely novel. In India, Section 52 of the Copyright Act, 1957, deals with the concept of fair use doctrine which specifies acts that do not constitute copyright infringement. It primarily establishes exceptions, including "fair dealing" for purposes like private use, criticism, and news reporting, and also covers reproductions for judicial proceedings, educational uses, and certain public performances.

The more creative the original work, the stronger the copyright protection. Since Studio Ghibli's films are highly creative works rather than factual compilations, this factor leans in favor of copyright holders. This case presents the legal issue of the “Idea-Expression Dichotomy”. This dichotomy presents the view that ideas themselves are not protectable under copyright laws but the expression of these ideas are as original creative works are copyrightable. As per Article 9(2) of the TRIPS Agreement, or Trade-Related Aspects of Intellectual Property Rights, a WTO agreement that sets minimum standards for intellectual property protection for all WTO member nations reads: “Copyright protection shall extend to expressions and not to ideas, procedures, methods of operation or mathematical concepts as such.” From this perspective, one can argue that studio Ghibli owns the copyright to its films and creations, but they may not own the copyright to the general idea of the animated art style.

One of the most contentious aspects of copyright law concerning AI-generated content is the amount and substantiality of the copyrighted material used in training datasets. This factor plays a crucial role in determining whether the use of copyrighted content constitutes fair use or infringement. The challenge arises from the fact that AI models do not copy specific works verbatim but instead digest and learn patterns from vast datasets, which may include protected works. Courts have yet to establish a concrete legal threshold for how much copyrighted material AI models can lawfully use in training. If an AI model is trained on thousands of Ghibli frames, does that constitute excessive copying? Courts have yet to determine a threshold for how much copyrighted material AI can legally use in training. Under U.S. copyright law, the third fair use factor, outlined in 17 U.S.C. § 107, examines both the quantity and the quality of the copyrighted work used. Courts often apply the qualitative vs. quantitative test to determine whether the portion taken is substantial enough to constitute infringement:

1. Quantitative Use: If an AI model is trained on thousands of frames from Ghibli films, does it amount to excessive copying? While AI does not store images in their original form, its ability to generate highly recognizable outputs that resemble Ghibli’s artistic style suggests that a substantial portion of the studio’s creative essence has been incorporated. Courts may look at the percentage of the dataset consisting of Ghibli frames to determine if the use is excessive.

2. Qualitative Use: Even if an AI model uses only a small portion of copyrighted material, it may still be considered infringement if the portion used is the “heart” of the original work. In Harper & Row v. Nation Enterprises, 471 U.S. 539 (1985), the US Supreme Court ruled that even a brief excerpt could be infringing if it captures the most expressive or essential elements of the work. Applying this principle, courts could argue that training AI on Ghibli’s unique animation style arguably the core of its artistic identity could constitute substantial copying, even if the dataset represents only a fraction of Ghibli’s overall works.

A critical factor in fair use determinations is whether AI-generated content competes with or diminishes the market for the original works. Studio Ghibli has a strong commercial market for its animation, including films, licensed merchandise, and commissioned artwork. If AI-generated Ghibli-style images replace the need for licensed artwork or official collaborations, this could constitute market harm, weighing against fair use. Additionally, the economic model of AI image generators where users can create derivative works without compensating the original artists raises concerns about unjust enrichment. In Dr. Seuss Enterprises v. ComicMix LLC (2020), it was ruled that a Dr. Seuss parody was not fair use because it replicated significant creative elements without adding enough transformation. In this case the visual style and textual structure were replicated to form a mash-up content. During adjudication it was held that the contents of the original work were copied and was rivaling the original content in the same market, thus was not a case of fair use. On similar lines, if AI models replicate Ghibli’s animation techniques without substantial creative contribution, they may be deemed infringing rather than transformative.

From the perspective of AI proponents and digital artists, the use of AI to create Ghibli-style artwork represents an evolution in artistic tools rather than a direct violation of intellectual property rights. They argue that just as human artists learn by studying the works of great masters, AI models "learn" by analyzing existing datasets. In this view, AI-generated content is akin to an advanced form of inspiration rather than outright copying. However, copyright advocates argue that AI-generated content lacks true creative intent and merely replicates styles without understanding or original expression. Unlike an artist influenced by Ghibli who creates something uniquely personal, AI models rely on pattern recognition to mimic an existing aesthetic. This leads to concerns that AI-generated art could undermine human creativity, devalue original works, and erode the incentive to create. Another critical concern is attribution. In traditional art, even when an artist is inspired by another’s style, they acknowledge their influences. AI-generated art, however, often lacks such transparency, making it difficult to determine whether it is ethically or legally appropriate to use. Several cases are under adjudication in various courts that address the use of copyrighted material in AI training, though none have definitively settled the issue. The case of Andersen vs Stability AI (2023) involves lawsuits from artists who claim their works were used without permission to train AI models. Similarly, Getty Images have filed lawsuits against AI developers who allegedly used their copyrighted images in datasets. These cases will likely set important legal precedents that determine whether AI-generated content is a derivative work (and thus subject to copyright) or a transformative creation (and thus protected under fair use). If courts rule against AI companies, developers may be required to obtain licensing agreements before training models on copyrighted material. Conversely, if AI-generated content is deemed legal, it could drastically reshape intellectual property laws and the creative industry.

The Ghibli trend encapsulates the larger debate over AI, copyright, and creativity in the digital age. While AI-generated content has opened exciting new avenues for artistic expression, it also raises significant ethical and legal concerns about intellectual property rights. The lack of clear legal guidance on using copyrighted material in AI training datasets creates a legal gray area, leaving artists, developers, and copyright holders uncertain about their rights, responsibilities, and protections. As AI technology continues to evolve, legal frameworks will need to adapt to address these challenges. Courts and policymakers must strike a balance between fostering innovation and protecting the rights of original creators. Until definitive legal precedents are set, the debate over whether AI-generated Ghibli-style art is a tribute or a violation will remain one of the most pressing questions in the intersection of art, technology and law.