In an Era Where AI Creates Actors, Where Does the Law Stand?
The Yum Hye-ran Unauthorized Duplication Case Exposes the System's Naked Reality

A scene from the AI movie <i>Geomchimwon</i>.

A scene from the AI movie Geomchimwon.

View original image

Recently, a short film produced with artificial intelligence (AI) was released on YouTube. The title was Geomchimwon. The main character closely resembled actress Yeom Hye-ran—not just her face, but also her voice, mannerisms, and body shape were intricately replicated. The producers claimed they had obtained the actress's permission. However, Yeom Hye-ran's side stated, "There was no such agreement," and demanded the video be deleted. The video was made private on the same day.


This is an incident that cannot be dismissed as a simple entertainment industry issue. Where technology has advanced first, we have seen institutional voids fully exposed. The problem goes beyond the infringement of portrait rights. The producers not only achieved a detailed replication, but also falsely asserted that they had received prior consent. In this way, they layered unauthorized duplication with fabricated consent. While similar to conventional deepfake controversies, this case delves even deeper.


The crux of the problem lies in the potential for the public to be misled into believing the actor actually appeared in the film. The technology has reached the point where it blurs the boundary between reality and fiction. Such induced confusion disrupts not only individual rights, but also market order. The very moment it becomes unclear whether an actor really participated, the traditional structures of labor and contracts become destabilized.


The institutional mechanisms to regulate this are insufficient. In Korea, individuals' images have been partially protected through portrait rights and the right of publicity, but these have largely been shaped through precedents. Considering the current level of generative AI technology, which can replicate not just faces and voices but also expressions and movements, it is difficult to regulate new forms of infringement under the existing legal system. In particular, when a replicated persona is commercially distributed, there is a lack of clear standards on how such cases qualify as rights violations.


Abroad, responses to similar problems have already become concrete. In 2023, the US actors' and broadcasters' union (SAG-AFTRA) reached an agreement with studios after a 118-day strike, incorporating consent and compensation principles for AI digital replication into their collective bargaining agreement. In December of the same year, the agreement was ratified with 78% approval from union members.


Legislative action has also been taken. In 2024, the state of Tennessee enacted the ELVIS Act, regulating unauthorized replication of identity elements, including voice, through state law. That same year, California signed bills regarding digital replication and AI transparency. In August 2024, the European Union's AI Act came into effect. Transparency obligations for generative content will be fully implemented from August this year. This series of developments shows a regulatory paradigm shift—from unconditional to conditional acceptance of technology.


Unlike these cases, simply extending existing rights through interpretation has clear limits. It is time not to revise, but to rewrite the rules. Introducing a concept like "digital portrait rights" to protect AI-generated faces, voices, and actions would expand coverage. For instances where confusion with a real person is possible, procedures requiring advance consent could be considered, regardless of commercial use. Furthermore, imposing at least basic verification duties on platform operators would help implement controls at the distribution stage. However, as such regulations may conflict with freedom of expression, it will be essential to carefully design the scope and standards in parallel debates.



Most important of all is a shift in awareness. For a long time, we have regarded a face as merely part of a person. In the age of AI, it no longer belongs solely to the body. It is replicated, modified, and distributed as data. Yet, people still perceive those faces as the person themselves. This disconnect is the starting point of the problem. Even when a face becomes consumable data, the responsibility and impact remain with human beings.


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

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing