The civil lawsuit pitting South Korea’s three major terrestrial broadcasters against Naver commenced, marking a landmark confrontation over generative AI learning practices.

On September 18, the Seoul Central District Court’s Civil Division 63 (Presiding Judge Lee Gyu-yeong) held the first hearing in the damages case brought against Naver by KBS, MBC, and SBS. The broadcasters accuse Naver of infringing their rights by using their news articles to train its generative AI, HyperCLOVA, without consent. Naver, meanwhile, counters that it possesses authorization under contractual agreements with the broadcasters regarding news content usage.

Earlier this year, in January, the broadcasters filed suit claiming Naver had unlawfully used their news articles to train both HyperCLOVA and HyperCLOVA X, violating copyright and the Unfair Competition Prevention Act. They demanded damages and an injunction against the use of their news content for AI learning.

Representatives for the broadcasters asserted, “The defendant (Naver), with a dominant market position, has invested vast sums in using the fundamental resource of news content without permission for its commercial AI products. At the core, this case is about the straightforward fact that news was copied and transmitted for generative AI without the explicit authorization of the plaintiffs.” Each broadcaster seeks damages amounting to 200 million KRW from Naver.

Naver disputed this claim, stating, “It is unclear which parts of the articles the plaintiffs seek to restrict, making it difficult to exercise our right of defense. The specific content being challenged must be identified.” Furthermore, Naver argued, “There are clear terms of service submitted by the plaintiffs, and news content was supplied per those contractual agreements.”

Naver also pointed to the Copyright Act, emphasizing that “news used for current affairs reporting is explicitly excluded from copyright protection,” and argued that the articles cited in the lawsuit are not protected works under the law, allowing for free use.

During the proceedings, the broadcasters submitted evidence indicating they had directly inquired whether their articles had been used to train Naver’s AI model and received responses from the AI suggesting in the affirmative.

In response, Naver questioned the credibility of such evidence, explaining, “HyperCLOVA X is itself prone to inaccuracies, and its answers often reflect the intent of the person asking the question. It is doubtful whether this material can serve as proof of AI learning.”

The presiding judge summarized, “The defendant holds that the actual copyrighted works at issue have not been specified, while the plaintiffs respond that, given the novel nature of this litigation, it is difficult to provide further detail.” The court emphasized the necessity for both sides to clarify the specific copyrighted materials allegedly infringed upon so that both the judges and the opposing party are aware.

The court further noted the contention over the news content provision agreements and the issue of whether current affairs reporting is exempt from copyright protection, asking both sides to submit additional written arguments.

Towards the end of the hearing, the broadcasters’ representatives stressed, “Globally, big tech companies such as ChatGPT’s developer OpenAI are moving towards properly licensing content from media organizations for AI development. According to Naver’s arguments, AI developers would be able to use other people’s copyrighted works at will without compensation.”

They added, “After hearing Naver’s defense today, it is ambiguous whether Naver did or did not use our news articles for AI learning. We ask that Naver clearly acknowledge if news content was used for training and present concrete evidence.”

In closing, the presiding judge requested that if Naver is able to specify which portions of the news content provided by the broadcasters were used for AI learning, they should review and clarify this scope.

The next hearing is scheduled for November 6, when the court expects to review both parties’ counterarguments and further statements.

Note “This article was translated from the original Korean version using AI assistance, and subsequently edited by a native-speaking journalist.”

Photo=Yonhap News