[The Bright and Dark Sides of AI Content] ③ AI Creation Gets Easier, but Regulations Lag... Bills Pile Up Awaiting National Assembly Review

AI Basic Act watermarking requirements unclear on scope and responsibility
Discussion of penalties for watermark tampering and measures to prevent facial identity theft
"Institutional safeguards needed as technology adoption accelerates"

Editor's NoteAI actors are being created without anyone standing in front of a camera, and influencers who do not exist in real life are gaining attention on social media. As AI lowers the barriers to content creation, it has opened up new opportunities for small-scale creators. While this has created an environment where content can be produced without professional equipment or a large budget, it has also increased the risk of unauthorized use of other people's faces, voices, and works. The Asia Business Daily examines how AI is changing the content market and explores ways to ensure that the potential of technology can coexist with people's rights.

Although artificial intelligence (AI) is being used to produce content, bills to strengthen labeling requirements for AI-generated content and improve protections against the unauthorized use of people's faces remain under review in the National Assembly. Experts have called for specific criteria defining the extent to which AI must be used before labeling is required, along with guidelines and self-regulation that can address a range of production and re-editing methods.

According to the National Assembly's Bill Information System on October 17, amendments to the AI Basic Act that would clarify labeling standards for AI-generated content and the parties responsible for labeling, as well as amendments to the Information and Communications Network Act that would prohibit the removal or alteration of labels and strengthen platforms' management responsibilities, have been proposed but have yet to clear the review stage. The Legislation Review Subcommittee of the Science, ICT, Broadcasting and Communications Committee discussed the related amendments on September 15 but decided to continue reviewing them, saying further consideration was needed.

Article 31 of the AI Basic Act, which took effect in January this year, requires AI businesses that provide generative AI or products and services using it to indicate that the output was created by AI. However, the scope and subjects of the labeling requirement are unclear, making it difficult to apply uniformly in practice. A key question is how to label content made entirely with AI versus content that has been partially re-edited using AI. Some have also pointed out that, despite the different roles played by companies that develop AI and those that use it to provide services, imposing the same labeling requirements on both leaves unclear how they should fulfill their responsibilities.

The proposed amendments seek to clarify the rules by distinguishing between labeling methods and the parties responsible. Rep. Kim Daesik's bill would require watermarks, such as codes or text, to be embedded in the output itself and would impose criminal penalties for damaging, forging, or altering them. Rep. Kim Wooyoung's bill would exempt uses of AI as an assistive tool, such as generating backgrounds or adjusting colors, while clarifying labeling standards for creative works. Rep. Ahn Cheolsoo's bill would assign responsibility for machine-readable labels and user-perceptible labels according to the respective roles of AI developers and businesses that use AI. Amendments to the Information and Communications Network Act proposed by Reps. Jo Incheol and Park Chunggwon would require distribution platforms, as well as creators, to take technical and managerial measures to support labeling and prevent labels from being tampered with.

Bills are also being pursued to restrict the unauthorized use of faces and voices in AI content production. Rep. Kim Seongwon is seeking to explicitly include digital replicas of names, likenesses, and voices among the interests protected under an amendment to the Unfair Competition Prevention Act. An amendment to the Popular Culture and Arts Industry Development Act proposed by Rep. Kim Jaewon would restrict the use, without consent, of digital imitations that reproduce the face, voice, speech patterns, or movements of a specific popular culture artist in commercial advertising or promotions, and require disclosure when such imitations are used.

AI content can be endlessly reworked... "Regulations cannot keep pace with technological advances"

Changes in the share of AI-generated content by platform, released by U.S. digital marketing firm First Page Sage. From 2021 to 2026, the share of AI-generated content on major social media platforms continued to rise.

Experts also said that specific criteria are needed to assess AI use based on its scope and the production process.

Lee Yongmin, a lawyer at Yulchon LLC, said, "If part of the content was produced by an outside contractor, the question arises as to whether the party using that content must also label it," adding, "Clear standards must be established for how extensively AI was used at each stage of production."

Park Hyungjun, a professor in the Department of Public Administration at Sungkyunkwan University, said, "The problem with AI content is that it is continually reworked," explaining that "it can be difficult to determine whether labeling is required when part of AI-generated content is excerpted and reused." He added, "Since it is difficult to regulate every piece of generated content by law, this needs to be supplemented with self-regulation that provides guidelines."

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