Korean Broadcasters Launch Copyright Lawsuit Against Naver Over AI Training Data

COMPANY / Reporter Paul Lee / 2025-09-18 23:57:31

Photo courtesy of Yonhap News

 

 

[Alpha Biz= Paul Lee] SEOUL, September 18, 2025 – South Korea’s three major terrestrial broadcasters, KBS, MBC, and SBS, have filed a lawsuit against Naver, alleging that the company unlawfully used their news content to train its generative AI model HyperCLOVA X without explicit permission. The first substantive hearing took place today at the Seoul Central District Court (Civil Division 63).



The broadcasters argue that Naver, leveraging its dominant market position, replicated and transmitted news articles without authorization, using them as training material to build a commercial AI product. Their claims include:



Copyright infringement

Unlawful use of data

Misappropriation of output

General tort liability under civil law



The plaintiffs are seeking KRW 200 million (approx. USD 145,000) in damages per broadcaster, in addition to injunctive relief.



The broadcasters also contend that Naver’s user agreement does not grant rights for AI-related services, and that Naver Cloud, a separate legal entity, had no authority to use the news data. “If the defendant’s argument is accepted, it would mean AI developers can freely exploit the byproducts and outcomes of others’ work without compensation,” their legal counsel argued in court.



Naver rejected these claims, asserting that its terms of service provide sufficient authorization for AI training. The company further cited Article 7(5) of Korea’s Copyright Act, which excludes “current news reporting” from copyright protection, arguing that the case differs from recent U.S. disputes over AI training with copyrighted materials.



Naver also questioned the sufficiency of the broadcasters’ evidence, noting that the prompts presented did not prove specific instances of copyright infringement. While the company did not explicitly confirm whether news articles were used for AI training, it stated in filings that any data processing involved “parameter-level training” rather than reproduction of copyrighted works.



The case is expected to set an important precedent in Korea regarding the intersection of copyright law and generative AI training practices.

 

 

 

 

Alphabiz Reporter Paul Lee(hoondork1977@alphabiz.co.kr)

주요기사

KT Confirms Additional Victims in Unauthorized Micro-Payment Case, Faces Criticism Over Initial Response
KFTC Grants Conditional Approval for Joint Venture Between Shinsegae Group and Alibaba Group
Supreme Court Rules in Favor of NTS on U.S. Unregistered Patent Royalties
Korea Zinc to Cancel All Treasury Shares by Year-End, Targeting Shareholder Return Ratio Above 200%
NextTrade to Suspend Trading of 66 Additional Stocks from September 22
뉴스댓글 >

SNS