“Ghiblification” Goes Global: AI’s Newest Copyright Controversy

“Ghiblification” Goes Global: AI’s Newest Copyright Controversy 

Popular memes (“distracted boyfriend”, “bro explaining” and the Turkish Olympic pistol shooter), historic images (Donald Trump’s defiant fist raise immediately after the assassination attempt and his heated meeting with Volodymyr Zelensky at the White House) and even famous movie trailers (Lord of the Rings trailer) were all subject to ghiblification by internet users. The trend garnered such attention and momentum that even official news outlets and government accounts jumped on the bandwagon to ghiblify images, joining the viral frenzy. Sam Altman, the CEO of OpenAI, encouraged the trend by changing his profile on X to a Ghibli-style portrait. 

Things took a turn for the worse when users started ghiblifying images of disasters and terrorist attacks such as the 9/11 image of the Twin Towers, which prompted many users to raise the issue of copyright infringement of Studio Ghibli’s works. 

Some even brought up Hayao Miyazaki’s comments on AI animation from 2016, in which he famously said, “I would never wish to incorporate this technology into my work at all” and “I strongly feel that this is an insult to life itself.” 

Copyright Controversy 1: Input Material for Training 

The law governing copyright in the UK is primarily set out in the Copyright, Designs and Patents Act 1988 (CDPA 1988). Copyright is a property right that recognises the intellectual creation of an author when a work is created. Copyright also arises automatically upon the creation of a work, provided it is original and not copied from another source. 

Copyright infringement occurs when someone uses the whole or a substantial part of the right holder’s work without the right holder’s permission and none of exceptions to copyright apply. A “substantial part” is not defined in copyright law but has been interpreted by the courts to mean a qualitatively significant part of a work even where this is not a large part of the work. 

Under UK law, if OpenAI has inputted whole or substantial parts of copyrighted material from Studio Ghibli (its movies and television shows) to train their AI model without the Studio’s permission, consent or a license to do so, then there could be a case of copyright infringement. However, if all OpenAI used as input material to train the model was fanart, then Studio Ghibli could have less of a case for copyright infringement here. 

Interestingly, the law in Japan, where Studio Ghibli is based, has specific provisions on generative AI. Article 30-4 of Japan’s revised Copyright Act came into effect on January 1, 2019, allowing broad rights to ingest copyrighted works for any type of information analysis, including for the purpose of training AI models. According to reports, Japan’s Minister of Education, Culture, Sports, Science and Technology, Keiko Nagoaka indicated AI companies in Japan can use “whatever they want” for AI training “regardless of whether it is for non-profit or commercial purposes”. This position led to Japan being called a “machine learning paradise.”  

However, with recent pushback and concerns from Japan-based content creators like other anime or manga developers, Japan’s Agency for Cultural Affairs clarified that digesting copyrighted content for “information analysis” (e.g. input training for ChatGPT for scholarly assessments) is allowed but the use for a “purpose of enjoyment of the thoughts expressed in the copyrighted work” is not allowed, meaning that if the intention is to output products that can be perceived as creative expressions of copyrighted works, including mimicking the style of specific creators without their consent, like the Ghiblification trend, then digesting copyrighted content is not allowed. 

Thus, in Japan, the legality of the ingestion of copyrighted material essentially appears to be tied to its intended generative AI output. 

Copyright Controversy 2 – Output Material and Artistic Style 

On Tuesday 25 March 2025, OpenAI expressed in a technical paper that the new tool would be taking a “conservative approach” in the way it mimics the aesthetics of individual living artists. While the tool has a refusal which triggers when a user attempts to generate an image in the style of a living artist, OpenAI mentioned that it “permits broader studio styles – which people have used to generate and share some truly delightful and inspired original fan creations.” 

In the UK, copyright protection is granted to specific categories of works, such as literary, dramatic and artistic works, provided they are original. Artistic style or an artistic idea itself is not protected under copyright law. 

There is speculation that OpenAI’s move to restrict the copying of the styles of living artists seem to be more due to the fact that individuals would get upset by the mimicry, rather than to prevent or mitigate any possible copyright infringement.  

However, by permitting broader studio styles, is OpenAI still on the hook for copyright infringement? Possibly, but the answers are not as black and white as the skillful prints on a page of the Spirited Away manga.  

As mentioned above, under UK law, a “substantial part” does not have to mean a large part of the original work being copied. If you freeze a frame in any of the Studio Ghibli films and compare them to outputs of the generative AI and see identical or similar elements in the output that are part of more specific and discrete elements of a Ghibli piece of art, then it is possible that OpenAI could have created works that resemble Studio Ghibli’s copyrighted products and perhaps copied the work of individual Studio artists. These artists could then bring a copyright claim. 

In Japan, there is specific law to determine copyright infringement in the output phase of generative AI. It seems that it is necessary to prove both “similarity” and a “reliance” on an existing copyrighted work. The “similarity” element follows the UK “substantial part” element closely, while the “reliance” element is shown when the creator of the new work is aware of the prior work. This means that if the AI user prompts ChatGPT to generate a specific Studio Ghibli copyrighted work by referencing it in a request prompt, the AI user may be found to be the infringer. However, if the AI user is not aware of an existing copyrighted work, and a copyright infringing output is generated, OpenAI could be found to be a contributory infringer. 

Generative AI and IP: Familiar Foes 

This is not the first high profile copyright issue that has proliferated the generative AI industry. Earlier this year, a court filing alleged that Meta CEO Mark Zuckerberg approved the company’s use of a notorious “shadow library”, LibGen, which contains more than 7.5 million books, to train Meta AI’s models. Many authors discovered that their works might have been used as input training material and a response letter was published claiming “clear infringement of copyright law”. A spokesperson from Meta claimed that the use of the material was exempt under “fair use” (The “fair dealing” exceptions in the UK are more limited than the “fair use” doctrine in the US, they are not interchangeable). The legal battle is ongoing. 

Conclusion 

Although Ghiblification might have started off as a fun trend on the internet, it has brought to light once again the struggle copyright holders face when Big Tech companies are able to use copyrighted data to feed their AI models without properly compensating right holders. As AI tools continue to improve, and countries start realising the desperate need to reconcile AI development with fair compensation to right holders, a solution needs to be found quickly.  

We are seeing a real surge in questions around copyright and intellectual property in the age of AI — from protecting AI-generated works to navigating cross-border rights. If you’re building an AI-driven product or wrestling with an IP conundrum of any kind, our team is here to help you cut through the complexity and find clear, practical solutions. Feel free to get in touch! 

Shennind Awat-Ranai 

Disclaimer: The above commentary is provided for general information purposes only and does not constitute legal advice. While KaurMaxwell is a firm of UK-qualified solicitors, we are not qualified to practise law in Japan. Accordingly, any references to Japanese copyright law are based on publicly available sources; we cannot confirm their accuracy or completeness. Readers should seek specific advice from a suitably qualified Japanese legal professional before relying on any such information.