Rival parties brace for deepfake flood as presidential election approaches

Home > National > Social Affairs

Rival parties brace for deepfake flood as presidential election approaches

Published: 21 Apr. 2025, 15:44 Updated: 22 Apr. 2025, 19:35

Audio report: written by reporters, read by AI

Members of Democratic Party candidate Lee Jae-myung’s election campaign team walk toward the Seoul Metropolitan Police Agency in central Seoul on April 16 to file a complaint, seeking charges of defamation and violations of the Public Official Election Act against the distributor of a deepfake video of candidate Lee. [NEWS1]

During the presidential election campaign, candidates have not only each other to fight but a third common enemy: Deepfakes.

With just weeks remaining until the June 3 presidential election, political parties are bracing for a wave of “deepfake negative campaigning,” as advances in deep learning technology make fabricating a candidate’s face or voice easier than ever.

Democratic Party (DP) leading candidate Lee Jae-myung declared a zero-tolerance rule against deepfakes.

“Reporting deepfakes is a must. We’ll respond to every single one as they appear,” is the stance of Lee's campaign team, which has adopted a no-leniency approach to smear campaigns using AI-generated content.

“We are actively searching for and analyzing posts right now,” said a member of Lee's legal support team on Sunday, adding, “False and manipulated information will not be tolerated.”

The strategy is to act preemptively and underscore the campaign’s transparency, even if it means that AI-generated content gets shared among other people as a result.

“We received a tip-off about an attempt to produce and circulate a deepfake video in which Lee verbally abuses his wife Kim Hye-kyung,” said Lee's campaign spokesperson, describing it as a fabricated video where Lee is depicted angrily confronting Kim as she returns home after questioning by investigators.

The rival People Power Party (PPP), on the other hand, is considering softer action on concerns that making a fuss would "just help spread it further and make it seem more real.”

In some deepfake videos circulating quietly on social media, PPP candidates are shown in fictional situations, with some scenes deemed explicit enough that staff worry any official response could backfire.

During internal discussions, opinions were split on filing a complaint in the candidate’s name or letting it pass quietly.

Screenshot of a deepfake YouTube video, reported to the Seoul Metropolitan Police Agency on Arpil 16 by Democratic Party candidate Lee Jae-myung’s campaign for alleged election law violations. [SCREEN CAPTURE]

“Our position is firm on pursuing legal action against defamation and spreading false information,” a party official said, remaining careful about other forms of content.

A dedicated attorney is monitoring content in real time, and they plan to respond collectively, the official said.

“With the recent boom in learning tools like ChatGPT, even people without specialized skills can now create synthetic videos,” Lee Seong-yeob, a professor at Korea University’s Graduate School of Management of Technology, said.

“Compared to three years ago, much more realistic AI-generated content is being used in fierce attacks in this deeply polarized election cycle.”

Campaigns have formed dedicated teams to monitor content and have begun filing complaints.

Following its “firm response to demonization” policy, Lee's campaign filed a complaint with the Seoul Metropolitan Police Agency on April 16 against 17 individuals, including conservative YouTuber Seong Chang-kyeong, for violating the Public Official Election Act.

One case includes a clip where a fabricated image of Lee Jae-myung is seen “giving orders to Democratic Party floor leader Park Chan-dae using profane language.” Another example is a YouTube video titled “The Republic of Great China” that shows Lee Jae-myung saying things like “Let’s ruin the country together” and “I’ve added a vasal state for Brother Xi Jinping.”

Many comments on the video note that the manipulation is blatant.

Still, as with the PPP case, political camps are grappling with how far to push their response. In an online environment dominated by polarized political YouTube channels, enforcement can backfire.

“Once a deepfake clip is identified, opposition YouTubers start using it as a marketing tool,” a DP aide said. “They just pay a fine and make more money off the views.”

“During the whole martial law incident, there were floods of deepfake videos of former President Yoon Suk Yeol, but only one complaint was actually filed,” a PPP official said. “We couldn’t file complaints for every case during the 2022 presidential election or last year’s general election either.”

Screenshot of deepfake videos of People Power Party candidate Han Dong-hoon and former President Yoon Suk-yeol, which have recently spread across social media platforms such as TikTok. [SCREEN CAPTURE]

Some observers say that with AI emerging as a major campaign issue this cycle, political parties are reluctant to call for heavy-handed regulation.

Although no one has yet been convicted, the revised election law that took effect in January last year includes a provision allowing for up to seven years in prison for violations involving deepfake videos. The National Election Commission formed a joint public-private response team on April 9, including prosecutors, police and platform operators like Naver and Kakao.

However, experts cautioned against blanket enforcement.

“Not all deepfakes involve defamatory content,” said Han Sang-hie, a professor at Konkuk University Law School. “During election seasons, which demand the highest possible protection for political speech, it’s not desirable for the state to impose sweeping regulations. A better approach would be to loosen restrictions while ensuring that damages are handled afterward through civil claims or lawsuits.”

Translated from the JoongAng Ilbo using generative AI and edited by Korea JoongAng Daily staff.

BY SHIM SAE-ROM [[email protected]]