How Does ChatGPT Use Generative AI to Imitate Ghibli Art Style? — A Legal Perspective on Portrait Rights and Intellectual Property Issues Behind Generative AI

In recent years, the rapid development of generative AI has enabled artificial intelligence to create stunning images. Among these technologies, ChatGPT and related AI image generation models can mimic different artistic styles, including the beloved Studio Ghibli art style. However, the application of such technology also involves legal issues related to portrait rights, intellectual property rights, and fair use, which warrant in-depth discussion.

How Does AI Imitate the Ghibli Art Style?

Although ChatGPT itself is primarily a natural language processing (NLP) model, it can integrate with Generative Adversarial Networks (GANs) or Diffusion Models to generate AI-created images. These technologies analyze large datasets to learn specific artistic styles, including color schemes, line work, and background details, enabling them to generate highly Ghibli-like visuals.

For example, when a user inputs a prompt such as "generate a town scene in the Ghibli style," AI will produce an image that aligns with Studio Ghibli's artistic characteristics. However, if this application involves elements protected by intellectual property rights, such as directly mimicking characters or scenes from Hayao Miyazaki's films, it may lead to copyright infringement issues.

Is an Artistic Style Protected by Copyright?

Under the "idea-expression dichotomy" in copyright law, only specific expressions are protected by copyright, while abstract ideas or concepts are not. This principle ensures that artistic styles, techniques, and themes remain outside the scope of copyright protection. For instance, a painter’s unique brushwork or color choices may be recognizable as their personal style, but if another artist creates a new work using a similar style, it would not constitute infringement[1].

In the 2003 case of Satava v. Lowry[2], the U.S. Court of Appeals for the Ninth Circuit ruled that merely imitating an artistic style (such as a particular type of sculpture) is insufficient to establish copyright infringement unless the work contains unique and protectable creative elements. Similarly, while the Ghibli style itself is a form of artistic expression and is not protected by copyright, AI-generated images that closely resemble copyrighted characters or scenes could still constitute infringement. Therefore, when using generative AI, it is crucial to avoid excessive similarity to protected works to mitigate legal risks.

Portrait Rights Issues

Another legal concern with AI is portrait rights. When AI is used to generate realistic human faces or transform real-life photos into a Ghibli-style image, it may violate an individual's portrait rights. For example, converting characters from The Legend of Zhen Huan into a Ghibli-style image, or adding fantastical effects to real actors while altering the storyline, could infringe upon the actors’ portrait rights.

In many jurisdictions, individuals have control over their likeness, and unauthorized use for commercial purposes may violate their rights. Unlike copyright, portrait rights are part of personality rights and do not fall under the fair use doctrine.

According to Article 18 of Taiwan Civil Code, individuals whose personality rights are infringed may request the court to cease the infringement, prevent potential violations, or, in certain circumstances, claim damages or moral compensation. Additionally, Article 195, Paragraph 1 of the Civil Code states:

"If a person unlawfully infringes on another's body, health, reputation, freedom, credit, privacy, or chastity, or otherwise seriously violates another’s personality rights, the injured party may claim damages even if no financial loss has occurred."

Whether a violation of portrait rights warrants compensation depends on whether the infringement is deemed "serious." Judicial precedent suggests that factors such as whether the victim is a public figure, the context of use, and the purpose of the usage should be considered. For instance, unauthorized use of a public figure’s likeness for commercial purposes is typically deemed a serious violation. Similarly, if a private individual’s image is misused in a manner that damages their reputation or is manipulated in a misleading way, it could also qualify as a serious infringement, warranting legal action[3].

Therefore, when AI-generated content involves real individuals' likenesses, obtaining proper authorization is crucial to avoid legal disputes.

Are AI-Generated Works Protected by Copyright?

While the Ghibli art style is not copyrighted, the characters created by Studio Ghibli are protected under copyright and trademark law. For example, generating an AI-created character that closely resembles Totoro could still constitute copyright infringement. To minimize legal risks, businesses or individuals looking to use AI-generated Ghibli-style content should consider legal alternatives, such as purchasing licensed assets, using open-source datasets, or collaborating with artists while ensuring proper post-processing to comply with legal regulations.

Regarding whether AI-generated works themselves are eligible for copyright protection, this issue depends on specific circumstances. According to Taiwan’s Intellectual Property Office, AI is neither a natural person nor a legal entity. Therefore, works created solely by AI are not considered "works" under copyright law and do not qualify for copyright protection. However, if a human actively participates in the creative process and the AI merely serves as a tool, the resulting work may be protected under the human author’s copyright[4].

Challenges in Proving AI-Generated Copyright Infringement

To claim copyright infringement, the plaintiff must prove that the alleged infringer had access to their work and that the infringing work is "substantially similar" to the original. However, AI-generated content is typically produced based on user prompts, making it difficult for an infringed party to prove that the AI model had prior exposure to their work—unless the original work is highly recognizable.

Similarly, in cases involving portrait rights violations, damage claims depend on the ability to demonstrate actual harm. Under Taiwanese law, compensation claims require proof of damages; if the plaintiff cannot prove how the unauthorized use of their likeness caused harm to their reputation or led to negative consequences, their claim may not succeed in court[5].

Conclusion: Leveraging Generative AI While Respecting Intellectual Property and Portrait Rights

Generative AI has brought exciting innovations to the intersection of creativity and technology, allowing ChatGPT and GAN-based models to replicate the Ghibli art style. However, when using these technologies, it is essential to be mindful of legal implications related to portrait rights and intellectual property rights to ensure compliance.

By adhering to fair use principles and obtaining proper authorization, we can enjoy the creative potential of AI while minimizing legal risks, fostering a balance between technology, innovation, and artistic integrity.