KISDI Holds First Meeting of the Technology and Norms Subcommittee of the AI Social Policy Forum (detail)

※ URL(Korean): https://www.kisdi.re.kr/bbs/view.do?bbsSn=115095&key=m2101113055776&pageIndex=1&sc=&sw=

KISDI Holds First Meeting of the Technology and Norms Subcommittee of the AI Social Policy Forum

Policy Options Discussed, Including Extending AI-Generated Content Labeling Requirements and Mandating the Retention of Training Data Usage Records

▲ Review of extending AI-generated content labeling requirements to the distribution stage**

▲ Discussion on the effectiveness of legally mandating the disclosure of training data usage records**

▲ General discussion with experts from academia, industry, and the legal sector**

Date and Time: August 14, 2026 (Friday), 14:00–16:00
Venue: HJ Business Center, Gwanghwamun, Seoul
Hosted by: Ministry of Science and ICT (MSIT)
Organized by: Korea Information Society Development Institute (KISDI)

The Korea Information Society Development Institute (KISDI, President Sangkyu Rhee), together with the Ministry of Science and ICT (MSIT), held the first meeting of the Technology and Norms Subcommittee of the AI Social Policy Forum on August 14 at the HJ Business Center in Gwanghwamun, Seoul.

Participants included Professor Sang-Wook Lee of Hanyang University, Chair of the AI Social Policy Forum; Professor Hyun-Kyung Kim of Seoul National University of Science and Technology, Chair of the Technology and Norms Subcommittee; and subcommittee members representing academia, industry, and the legal sector.

The meeting addressed two key issues: whether AI-generated content labeling requirements should be extended to the distribution stage, and whether the retention of training data usage lists or records should be mandated by law. It provided the first opportunity to share the forum's deliberations to date and gather opinions on the direction of policy recommendations.

The Framework Act on the Development of Artificial Intelligence and Establishment of a Foundation of Trust (AI Basic Act), which took effect on January 22, 2026, requires AI business operators to label content generated using generative AI. However, the law applies to businesses that develop AI or provide AI-based products and services. It does not impose labeling obligations on online platforms or other entities involved in distributing AI-generated content, where users encounter such content.

Disputes over training data have also continued. South Korea's three major terrestrial broadcasters filed copyright infringement lawsuits against Naver in January 2025 and OpenAI in February 2026, alleging unauthorized use of news content for AI training. The Korea Newspaper Association has also filed a related complaint with the Korea Fair Trade Commission. However, the lack of a clear means to determine which data were used for training has been identified as an obstacle to protecting the rights of copyright holders.

The meeting began with a presentation by Jae-Kyung Shin, a principal investigator at KISDI, titled "Operational Plan for the AI Social Policy Forum." The presentation outlined the forum's overall operational direction and the background to the discussions of the Technology and Norms Subcommittee. It was followed by presentations on the two main policy issues.

The first presentation, titled "Review of Expanding the Scope of Entities Subject to AI-Generated Content Labeling Requirements," was delivered by Eunji Son, Visiting Professor at Sookmyung Women's University.

Current labeling requirements primarily apply to businesses that develop or provide AI. However, users encounter AI-generated content through search engines, online platforms, and individual services. A key policy question is therefore how far the scope of entities subject to labeling obligations should extend to ensure their effectiveness. Drawing on the European Union's AI Act and Digital Services Act, Professor Son examined approaches to developing a digital regulatory framework covering both the generation and distribution stages.

The second presentation, titled "Trade Secret Protection of Training Data Usage Records and the Effectiveness of Mandating Their Disclosure by Law," was delivered by Ilhyun Park, an attorney at Yulchon LLC.

Information about which data were used for AI training may constitute trade secrets that businesses seek to protect. At the same time, such information may provide creators with virtually the only means of determining whether their copyrighted works were used. Drawing on domestic and international cases, Attorney Park examined the effectiveness of legally mandating disclosure and possible approaches to regulatory design that account for these competing interests.

A general discussion followed, moderated by Professor Hyun-Kyung Kim of Seoul National University of Science and Technology, Chair of the Technology and Norms Subcommittee. Experts from academia, industry, and the legal sector discussed policy approaches addressing both the diffusion of AI technologies and the protection of rights.

Based on the opinions presented at the meeting, KISDI plans to continue reviewing policy issues related to AI-generated content labeling and copyright in training data. Through further subcommittee discussions and stakeholder consultations, KISDI will develop policy response measures that address these issues effectively.

KISDI News

KISDI Holds First Meeting of the Technology and Norms Subcommittee of the AI Social Policy Forum