[AI Chat] Big AI with Free Data... Will Copyright Fees Be Charged?

Novel-writing and Music-producing Generative AI... Copyright System Remains a 'Blank'
Controversy Over Unauthorized Use of Training Data... AI Copyright Issues Emerge
Government to Establish Digital Rights Charter Including AI Copyright by September

# Recently, new songs performed by popular American rapper Drake and singer The Weeknd have climbed to the top of music charts. However, the reality was that these songs were created by AI (artificial intelligence) synthesizing their voices convincingly. Drake's side issued a strong warning regarding copyright issues, and platforms promptly removed the tracks.

With the rapid rise of generative AI starting from ChatGPT, such cases are pouring in. The situation is too complex and impactful to be dismissed as a mere incident. Questions arise such as whether Drake can demand payment for the AI-generated use of his voice, or conversely, whether AI can receive royalties for the music. In South Korea, there is no answer to these questions because there is no law regarding AI copyrights.

The IT industry unanimously agrees that establishing a system for AI copyrights is urgent. An environment where legal risks cannot be anticipated could become an obstacle to industrial development. On the other hand, opposition voices from copyright holders in publishing, music, and other industries are strong, making the path to legislation long.

The Rapid Rise of Generative AI... Copyright Discussions Lagging Behind

The issues are mainly twofold: ▲ whether AI can use training data without permission, and ▲ whether the outputs produced by AI can be recognized as copyrighted works. Under current copyright law, if the purpose of data use falls under so-called fair use (public interest), it can be used without permission or compensation. This applies to uses such as school education or court trials. AI does not fall under this category but has used data for training under the premise of developing new technology for public interest. However, as companies began monetizing generative AI, the peace ended. Voices demanding that AI must obtain permission or pay fees to use data are growing louder. The outputs produced by AI are also outside the scope of the law. Current copyright law protects only creative works made by humans.

Discussions on institutionalization are still in their infancy. In February, the Ministry of Culture, Sports and Tourism launched the 'AI Copyright Law System Improvement Working Group,' which includes experts from various fields. The third meeting was held on the 24th of last month to share related issues. A similar consultative body operated in 2021 but was reinitiated this year. The situation changed as generative AI surged after ChatGPT appeared in November last year. The Ministry plans to create AI copyright guidelines incorporating the working group's opinions by September this year. Including this, the government announced plans to establish the 'Pan-Government Digital Bill of Rights,' a new digital order for the AI era.

Legislative movements in the National Assembly are also slow. A bill allowing AI to learn data without copyright issues was proposed but has seen no progress. A comprehensive revision of the copyright law has been dormant for over two years, and a partial amendment bill introduced last October remains pending in the standing committee. The committee, in its review report, stated, "To harmonize the protection of copyright holders' interests and the activation of the data industry, various legislative cases and opinions must be discussed in depth," thus deferring judgment.

"No Data Means an Empty Shell" vs. "Shaking the Copyright Industry"

The IT industry desires rapid institutionalization. Without legal stability, it is difficult to accelerate technology development or market expansion. Especially, data is essential fuel for AI development, but there are concerns that legal interpretations could cause issues under copyright law, personal information protection law, and others.

A CEO of an AI startup said, "It is already tough to keep up with capital battles and technological gaps, but if the system lags behind as well, domestic companies will only bear legal problems," adding, "It is necessary to clearly define what should not be done first." Professor Jung Sang-jo of Seoul National University Law School also emphasized at the seminar 'Copyright Issues and Challenges of Generative AI' on the 24th, "AI is an empty shell without data," and "To minimize uncertainty in the data war, copyright law revision is necessary."

On the other hand, opposition voices are also strong. They argue that granting exceptions to AI could undermine the foundation of copyright. There are concerns that systems designed to foster the AI industry could shrink other industries such as publishing and music. Accordingly, the Korea Music Copyright Association, the largest music copyright organization in South Korea, expressed opposition to the copyright law amendment bill. The Korea Press Foundation also submitted an opinion that the amendment could infringe on news copyrights of media companies.

Professor Choi Kyung-jin of Gachon University Law Department (President of the Korean Artificial Intelligence Law Association) pointed out, "There could be conflicts between the IT industry and cultural industries, so social discussion is necessary," adding, "Since ChatGPT has only recently appeared, more cases need to accumulate for institutionalization."

This content was produced with the assistance of AI translation services.