[Essential Intellectual Property] The 'Thing' That Trips Up AI... Even the 'Google Challenger' Couldn't Escape It

AI Industry Faces Dozens of Copyright Lawsuits Worldwide
Key Issue: Whether Fair Use Applies... No Precedents in Korea
"Secure Usage Rights from Copyright Holders to Prevent Disputes"

Companies utilizing generative artificial intelligence (AI) technology are facing difficulties due to ‘copyright’ issues. Recently, a former OpenAI researcher revealed that the company developed ChatGPT in violation of copyright law. AI company ‘Perplexity,’ in which SK Telecom invested 10 million dollars (13.7 billion KRW), was also sued for copyright infringement by a major media company.

‘Copyright’ Emerging as a Landmine in the AI Industry

As AI technology rapidly advances, copyright issues remain one of the challenges to be resolved. Overseas, dozens of lawsuits have been filed across various AI industry sectors such as language, image, music, and code generation. The main points of contention in lawsuits against AI operators are whether AI unlawfully learned copyrighted works or whether AI-generated outputs infringe copyrights.

It is no exaggeration to say that AI companies are at war with content creators. Getty Images filed a lawsuit against ‘Stability AI’ for unlawfully learning millions of images owned by the company, and The New York Times filed copyright infringement lawsuits against OpenAI, Microsoft, and Perplexity, known as ‘Google’s rival.’ The image generation AI ‘Midjourney’ is also currently facing lawsuits from multiple artists.

Is AI’s Learning of Copyrighted Works ‘Fair Use’?

Under copyright law, unauthorized use of another’s copyrighted work constitutes copyright infringement. However, it is not considered infringement if it falls under the exception of ‘fair use.’ Fair use must be evaluated by considering ▲the purpose and character of the use ▲the type and use of the copyrighted work ▲whether it unfairly harms the author’s interests ▲and the impact on the market.

Aravind Srinivas, co-founder and CEO of AI-based search company 'Perplexity,' held a press conference at the SK Telecom headquarters in Seoul last September. Recently, Perplexity has been involved in a copyright dispute over unauthorized use of content. Photo by Younghan Heo younghan@

Some argue that if existing ideas or content are not directly copied but used in new and innovative ways to enhance social benefits, it qualifies as fair use. A representative example of fair use is the ‘Google Books Library Project.’ This large-scale digital library project, started by Google in 2004, aims to digitize books worldwide for online search and viewing. Although many copyright holders filed lawsuits claiming unauthorized scanning and digitization of books, the project was recognized as fair use for improving access to information and promoting research and knowledge creation.

However, opinions in academia are divided on whether AI’s learning and use of copyrighted works qualify as fair use. There are no precedents yet in South Korea. Suchir Balaji, who worked as a researcher at OpenAI for about four years, revealed that he left the company because he believed that developing ChatGPT without considering copyright could cause more harm than benefit to society.

Copyright Commission: "Disputes Should Be Prevented in Advance"

In South Korea, there is a high possibility of lawsuits similar to overseas cases. However, lengthy court battles that consume time and money are likely to become a ‘no-win situation.’

The Korea Copyright Commission, through its ‘Generative AI Copyright Guide,’ recommended that "AI operators should, as much as possible, secure usage rights in advance from copyright holders through appropriate compensation or other methods, even if the use is for AI training, to prevent potential disputes."

Just because a work is publicly available on websites, blogs, or social networking services (SNS) does not mean it can be used without the copyright holder’s permission. AI operators need to be mindful of this when using copyrighted works. The Copyright Commission urged operators to specify details such as the purpose, scope, and duration of use when signing contracts with copyright holders.