President Lee: "Deepfakes Without Labels May Serve as ‘Real Evidence’... Distribution Stage Must Also Be Regulated"

President Lee Presides Over Ministry Work Reports
Visible Labeling Required for AI Content Altering Real Individuals
"Distributors Who Remove or Tamper With Labels Escape Regulation...Countermeasures Needed"
On Secondary Harm on Global Plat

On July 16, President Lee Jae-myung stressed that deepfake videos and images, which are almost indistinguishable from real people, could be accepted as evidential proof if distributed without clear labeling. He called for stronger regulations not only during the production, but also throughout the distribution phase.

President Lee Jae-myung is listening to the responses of the attending agency heads during the ministry work report held at the Blue House Yeongbingwan on July 16, 2026. Photo by Blue House Press Photographers Team, Yonhap News Agency

During the Ministry of Broadcasting, Media and Communications Committee's work report at the Blue House State Guesthouse, President Lee stated, "Nowadays, it is difficult to distinguish between AI-generated and real footage—it is remarkably similar." He added, "If content is not clearly labeled, it can serve as a form of evidence for people, making them believe, 'This must be real.'

He continued, "Without explicit labeling, it can lead to serious misunderstandings, and if exploited, it can have very grave consequences." He emphasized, "We must prepare countermeasures, as the risks associated with AI-generated content have already become a reality."

According to the current Basic Artificial Intelligence Act, videos and images created with generative AI must be marked as AI-generated. In particular, deepfakes—generative AI content that manipulates images of real people or actual phenomena—must be visibly identified so that users can instantly recognize them. Violating the labeling obligation can result in fines of up to 30 million won, but these enforcement regulations have currently been postponed for one year.

President Lee also specifically inquired about the legal definition and scope of deepfakes. The government explained that it interprets AI-generated content that manipulates real people or events as deepfakes, clearly distinguishing this from pure film or artistic works.

Citing examples of cases where a woman's face or body is superimposed onto explicit content, President Lee pointed out, "If such material is not labeled, it is subject to sanctions under AI-related laws in addition to general criminal law. However, enforcement is currently postponed, and the fines are limited to 30 million won or less."

He particularly highlighted the issue that current laws require labeling only from AI model developers and service providers, but do not directly regulate regular distributors who erase, crop, or redistribute content without the original label.

President Lee asked, "If a person only uses the unlabeled portions of an AI-marked video, does the labeling obligation not apply to them?" A government official responded, "According to the current Basic Artificial Intelligence Act, individual distributors and personal users on platforms like YouTube are not subject to regulation."

When the Broadcasting, Media and Communications Committee explained that a bill to regulate distributors who damage or delete labels has been proposed to the National Assembly, President Lee said, "It’s not enough to merely say 'We are watching to see what happens in the Assembly.' The agency must set a clear direction and actively seek cooperation."

The point is that even if creators attach labels, regulations can become ineffective if these are removed or edited during distribution. Therefore, the government should institutionalize responsibilities extending to the distribution stage.

The repeated distribution of deepfake victim posts on global platforms was also raised as a problem. The public participation group suggested that while secondary harm continues to target victims of deepfakes through portals, YouTube, and X (formerly Twitter), actual deletion of reported posts is extremely rare, so an effective reporting and remedy system is necessary.

President Lee pointed out that, for domestic portals, posts are blocked first upon defamation or rights violation reports from victims, then subject to an appeal process. In contrast, on overseas platforms, identity verification and takedown procedures are not properly enforced.

When the Broadcasting, Media and Communications Committee cited trade issues, technical workarounds, and budgetary constraints as realistic limitations in regulating global platforms, President Lee replied, "That does not mean we can simply do nothing. Enormous harm is occurring on the ground."

He went on to emphasize, "Explaining why something is impossible for this or that reason is of little value. What matters is for the government to consider how these harms can be prevented or minimized. That is our obligation."

On the 16th, at the Blue House State Guesthouse, during the ministry work report session chaired by President Lee Jae-myung, members of the public participation group raised their hands to request the floor. July 16, 2026.
Photo by Blue House Press Photographers Team, Yonhap News Agency

The Broadcasting, Media and Communications Committee stated that it is operating an integrated support team for victims of digital sex crimes and an inter-ministerial task force for AI-related crimes, and announced plans to strengthen measures against the dissemination of illegal and harmful content via global platforms.

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