If there are students who admire the works of master animators and recreate their personal and family photos in the master's artistic style, turning them into animated versions, most people would likely view this positively and not think about whether it constitutes copyright infringement.
If there are students who admire the works of master animators and recreate their personal and family photos in the master's artistic style, turning them into animated versions, most people would likely view this positively and not think about whether it constitutes copyright infringement.
If a student photocopies from a book of collection of a master’s works borrowed from the library in order to reference it and practice the master's style at any time, even though the final animated personal and family photos do not copy any specific work from the book, does it constitute copyright infringement? If the student photocopies from a book of collection of a master’s works that they purchased themselves for ease of use and study—does that make any difference in the determination? And what if we substitute the student with AI?
The copyright holder has the exclusive right to reproduce and adapt their work. Copying or modifying a master’s works without permission would infringe on the master’s copyright. However, copyright protection only extends to specific, tangible works—not to the ideas, concepts, or principles conveyed by a work. A master’s artistic style is considered an idea or concept, so recreating personal and family photos in an animation-inspired version that imitates a master’s style does not infringe the master’s copyright. That said, if the imitated work closely resembles a particular character from the master's works, it may still constitute reproduction or adaptation and thus infringe the master's copyright.
Photocopying is a form of reproduction. Whether the collection book of a master’s works was borrowed from a library or purchased by a person, photocopying it without the master’s permission may constitute copyright infringement. However, if only a portion is copied and it is for personal use—not made public or sold—it may fall under the protection of fair use. Fair use is explicitly defined by law as the use of another’s work within specific boundaries, which, even without authorization, does not constitute copyright infringement. This is typically permitted for reasons such as education, journalism, public interest, or the promotion of innovation.
Currently, copyright law does not have separate rules specifically for AI. If AI imitates a master’s style to generate animation-styled personal and family photos, and the resulting images do not closely resemble any specific character from the master’s works, then no copyright infringement occurs. The key difference lies in the learning process. Human learning involves memory in the brain; storing a master’s work in one’s brain is not considered reproduction or copying and does not infringe copyright. However, training AI requires inputting large volumes of works into a database, which does involve reproduction of the works. If this is done without permission from the copyright holder, it may constitute copyright infringement.
Whether the process of training AI could fall under the protection of fair use has become a hot topic in copyright law. Taiwan's Intellectual Property Office (TIPO) has pointed out that if the works of artists or illustrators possess originality and creativity, they are considered artistic works protected under the Copyright Act. If an AI art-generation tool connects to Internet resources and uses algorithms to "learn" from such works online, it may involve the reproduction of others' copyrighted works. Unless such use falls within the scope of fair use, the consent or authorization of the copyright holder should first be obtained. However, whether fair use can be asserted needs to be assessed on a case-by-case basis, and there is no fixed standard.
Several lawsuits have already been filed overseas by copyright holders against AI developers, claiming that using copyrighted works without permission to train AI models constitutes copyright infringement. These cases include The New York Times v. OpenAI, Microsoft & Getty Images v. Stability AI, and various artists v. Stability AI & Midjourney. In all of these cases, AI developers have invoked fair use as a defense against infringement claims.
As AI becomes increasingly widespread, integrated into daily life, and commercialized, the world is paying close attention to how these rulings will clarify the boundaries of fair use in the context of AI.
This article was published in the Expert’s Commentary Column of the Commercial Times. https://www.ctee.com.tw/news/20250423700123-431305