Naver Copyright Controversy Puts Pressure on Seoul to Clarify AI Training Laws (Image courtesy of Yonhap)
SEOUL, Oct. 13 (Korea Bizwire) — South Korean media organizations are preparing multi-billion-won lawsuits against Naver Corp., alleging the tech giant used news content without permission to train its artificial intelligence (AI) models, including the large language model HyperCLOVA X.
According to lawmaker Choi Soo-jin of the National Assembly’s Science, ICT, Broadcasting and Communications Committee, the Korea Broadcasting Association filed a lawsuit earlier this year against Naver and its cloud subsidiary, seeking 6 billion won (US$4.3 million) in damages — 2 billion won each for the nation’s three major terrestrial broadcasters. The group is expected to demand hundreds of billions of won in additional compensation as the case develops.
The Korea Newspaper Association also filed a complaint with the Fair Trade Commission in April, accusing Naver of exploiting its dominant market position to use news articles as AI training data without consent. Individual media outlets are expected to pursue separate damage claims once the regulator’s review concludes.
Court filings from the broadcasting association suggest that news content accounted for roughly 13 percent of Naver’s AI training dataset, which included blogs, online communities, comments, Wikipedia entries, and linguistic corpora from the National Institute of the Korean Language. The group argued that Naver “acknowledged using news content” in developing its AI services but never obtained permission from copyright holders.
Naver’s HyperCLOVA (Image courtesy of Naver)
The newspaper association further accused Naver of copying, summarizing, and restructuring key parts of articles through its AI-based products — including “Q:” and “AI Briefing” — and warned that AI-generated summaries have already led to distorted reporting and omitted key details.
Meanwhile, critics say the Ministry of Science and ICT, responsible for AI policy, has remained largely passive as copyright disputes escalate. When questioned about unauthorized use of news in AI training, the ministry responded only that “a balanced approach is needed to protect copyright holders while fostering AI innovation.”
Other ministries have taken a more proactive stance. The Ministry of SMEs and Startups, for instance, recently proposed legislation exempting small businesses from copyright liability in AI training.
Abroad, legal frameworks are clearer: the European Union broadly allows text and data mining unless explicitly prohibited by rights holders, Japan’s 2018 copyright reform permits both research and commercial use, and the United States applies the “fair use” doctrine to data training when deemed to serve public interest.
“Copyright conflicts in the AI industry are already turning into full-scale legal battles,” Choi said. “The government must act swiftly to establish exemption criteria and fair compensation systems for copyright holders before the issue spins out of control.”
Kevin Lee (kevinlee@koreabizwire.com)
