7 May 2025
The founder of Studio Ghibli, Hayao Miyazaki is famous for producing countless great anime films. His most notable works include “Spirited Away,” “My Neighbor Totoro,”, “Castle in the Sky”, “Ponyo,” and most recently, “The Boy and the Heron”. Hayao Miyazaki takes several years to produce the films due to his meticulous, hand-drawn animation style and his perfectionism.
Recently, OpenAI introduced its new image-generation tool for ChatGPT i.e. GPT-4o which allows users to render images into different styles based on the user’s request – for free and in a matter of seconds. The new tool has become immensely popular, with many people sharing AI generated images in various artistic styles such as Studio Ghibli style, Lego style and many more on social media.
This trend has raised broader discussions about the ethical concerns of AI generated art, intellectual property issues and the future livelihoods of animators and artists. The copyright issues in respect of AI generated images have sparked widespread condemnation online.
AI generated images come with many legal uncertainties, especially in the realm of intellectual property. Questions remain unresolved: Does copyright law apply to AI-generated works? Who owns the content created by generative AI— the platform, the user, or neither? And most importantly, can such AI generated image infringe existing intellectual property rights?
DOES COPYRIGHT LAW APPLY TO AI-GENERATED IMAGES AND WHO OWNS THE IMAGES?
Section 10 of the Copyright Act 1987 (“the Act”) sets out three conditions for a work to qualify for copyright protection.
Firstly, copyright is granted to “qualified persons,” which includes Malaysian citizens, permanent residents, or bodies corporate. This definition however does not seem to be extend to AI, which means that a work generated solely by AI would not satisfy this requirement. Secondly, the work was first published in Malaysia. Thirdly, the work was made in Malaysia.
The Malaysian High Court ruled in the case of Chuah Aik King (sole proprietor of Sykt B Three Technology) v Keydonesoft Sdn Bhd [2019] 8 MLJ 515 that the plaintiff has to fulfil one of the requirements in Section 10 for a work to be qualified copyright protection. If such criteria apply in respect of AI generated work, it can be argued that such work, if made or first published in Malaysia, may qualify for copyright protection, even if it was not created by a Malaysian ‘qualified person”.
Section 26(1) of the Act however provides that copyright conferred in Section 10 shall vest initially in the “author” of the work. The Act defines an “author” as a natural person who creates the work. This creates a significant hurdle, as AI is not a natural person and therefore cannot be considered an author under the current legal framework.
Although Section 26(4) of the Act provides legal presumptions for situations where the authorship or identity of the author is unclear, the rule on presumed authorship in this provision assumes that the ‘author’ is a natural person.
It should also be noted that in Malaysia, copyright protection for artistic works generally lasts for the author’s lifetime and an additional 50 years after their death. Since current laws only recognise human authorship, this potentially leads to indefinite protection for AI-generated works.
Arguably, if a human plays a meaningful creative role in the case of AI-generated work for example making specific prompts or making edits—the work may be eligible for copyright protection, with the human recognised as the author and possibly owns the work.
CAN AI GENERATED IMAGES INFRINGE EXISTING INTELLECTUAL PROPERTY RIGHTS?
The three elements required for establishing copyright infringement are set out in the Malaysian High Court case of Megnaway Enterprise Sdn Bhd v. Soon Lian Hock (No 2) [2009] 8 CLJ 130. For purposes of direct infringement, it must be established that:
- there is sufficient objective similarity between the original work or a substantial part thereof, and the infringing copy;
- there is a causal connection between the original work and the infringing copy, that is, the infringing copy must have been copied from the original work, whether directly or indirectly; and
- what has been infringed must constitute a substantial part of the original work.
Section 7(2A) of the Act clearly provides copyright protection shall not extend to any idea, procedure, method of operation or mathematical concept as such. The rationale behind this is that copyright law is meant to protect the expression of ideas rather than the ideas themselves. As such, If AI merely imitates a general style without copying substantial part of the protected work, it is unlikely to be considered infringement. Therefore, an individual merely posting his self-portrait generated by AI that mimics an artistic style without reproducing protected works is unlikely to face the risk of infringement.
On the other hand, if a person using AI-generated images for commercial purposes such as ads, packaging, or merchandise, he may face a higher risk of IP infringement as he doing so could mislead consumers into thinking that he is the official content producer and potentially harm the owner’s licensing market.
As an important viewpoint to consider, I asked ChatGPT for its view on the intellectual property concerns surrounding Studio Ghibli-style images generated by AI. ChatGPT answered as follows:
“When I generate an image “in the style of Studio Ghibli,” I’m not directly copying any specific Ghibli artwork or asset (like a character, scene, or exact frame), but I am mimicking their distinctive style. That raises a key issue:
Legally, it may not be considered direct copyright infringement unless the output is substantially similar to a protected work (like a Totoro character or Spirited Away scene). But…
Ethically and potentially under trademark or unfair competition laws, it could still be seen as infringing on Ghibli’s brand or artistic identity, especially if the generated content confuses people into thinking it’s official or affiliated.”
Interestingly, ChatGPT seems to agree that the legality of mimicking an artist’s style with AI remains a grey area.
In addition, the use of copyrighted images to train the AI to produce generated images is undoubtedly a legal or ethical issue as it is unclear if the AI companies have infringed intellectual property rights by using the copyrighted images as input. Perhaps, the AI companies could rely on exception and limitations to copyright protection under the Act such as fair use which allows the use of copyrighted material for purposes such as criticism, comment, news reporting, research, and education, without licence or permission from the copyright holder. However, there is no known precedents in Malaysia have been set.
CONCLUSION
In Malaysia, the authorship of creative works and the invention of new technologies are attributed exclusively to human. As AI generated image becomes increasingly common, current copyright and intellectual property laws must adapt to meet emerging challenges. Until clearer regulations emerge, AI companies, artists, and copyright holders should continue to navigate this uncertain area of law.
The complex issue of ‘originality’ under copyright law is not addressed in this article, but it is certainly a topic worth exploring in future discussions.
Written by Elaine Chan Yu Jing, Associate
This article is intended to provide general information and does not constitute legal advice. If you require further clarification and assistance, please feel free to contact us at info-cpkhon@cpkhon.com.
