TAIPEI (Taiwan News) — The Intellectual Property Office said it is considering fair use limitations for generative AI after a ChatGPT feature allowing users to create Studio Ghibli-style images went viral.

In a statement to CNA on Monday, the office said it must clarify whether generative AI that uses existing works could fall under fair use. It recommended that developers obtain permission from rights holders, adding that copyright infringement or plagiarism will be assessed on a case-by-case basis.

The office noted that copyright laws protect specific expressions of ideas but not artistic styles themselves. As such, while individual works are protected, AI-generated imitations that mimic general styles may not constitute copyright infringement.

It said that any disputes arising from such cases will likely need to be resolved through legal proceedings. The office also pointed out that some AI developers have imposed self-regulations to prevent users from copying identifiable artistic styles.

Officials said the office began researching the legal implications of generative AI following the release of ChatGPT in 2022 and plans to continue holding seminars and monitoring international developments. Because intellectual property issues are global in nature, finding a fair balance between the rights of creators and the development of the AI industry will require careful analysis, it said.

After launching the Studio Ghibli-inspired feature, ChatGPT said it would take a “conservative approach” to replicating artists' work, according to the Associated Press. The AI system will block attempts to mimic the style of living artists but allow broader studio-inspired styles.

Studio Ghibli did not comment on the matter, per AP. However, in 2016, studio founder Miyazaki Hayao said he would never use AI in his work.