laitimes

South Koreans digitize presidential candidates with deepfake and accept Q&A online

Reports from the Heart of the Machine

Machine Heart Editorial Department

"AI Yoon" is the first official "candidate" to synthesize using deepfake technology.

Recently, DeepFake technology was used in the South Korean election, constructing the world's first official deepfake political candidate.

The digital man, called "AI Yoon," is modeled after Yoon Suk-yeol, a candidate for the National Power Party. Using Yin's 20 hours of audio and video footage, as well as more than 3,000 sentences recorded specifically for researchers, a local deepfake technology company created the avatar AI Yoon and quickly became popular on the Internet.

Previously, AI technology has been used in elections, such as New York City mayoral candidate Andrew Yang, who ran in the virtual world. But the creators of "AI Yoon" believe it is the world's first official deepfake candidate.

At first glance, "AI Yoon" could be mistaken for a real candidate, with neatly combed black hair and a formal suit that looks almost exactly like Yoon Seok-yue himself. But to appeal to younger voters, the researchers asked "AI Yoon" to use some witty language and quips. Since its debut on January 1, it has attracted millions of views.

South Koreans digitize presidential candidates with deepfake and accept Q&A online

In fact, what AI Yoon expressed was written by the campaign team, not the candidate himself.

Kim Myuhng-joo, a professor of information security at Seoul Women's University, told local media that the script AI Yoon used to attend the event borrowed from the language used in online games. "AI Yoon read directly the script written by its creators," she says.

Baik Kyeong-hoon, head of the research and development team of "AI Yoon", said: Yin Xiyue's words often said in the real world are also well reflected in AI Yoon.

South Koreans digitize presidential candidates with deepfake and accept Q&A online

South Korea's National Power Party candidate Yoon Seok-wook.

Ai Yoon's rapid rise in popularity has boosted the popularity of candidate Yoon Seok-yue. Among voters in their 20s, Yoon Ise-yue is ahead of rival Lee Jae-ming.

This has a lot to do with AI Yoon's language style and its R&D team. AI Yoon creator Baik and two other technical members are the youngest employees in Yoon's campaign. Winning the votes of young voters may be one of their intentions.

South Koreans digitize presidential candidates with deepfake and accept Q&A online

The team generated responses to AI Yoon's brainstorming Q&A, which could take just 30 minutes. This is in stark contrast to the rhetoric that is often carefully polished in public policy debates. Baik believes that it is very easy to create large amounts of content using deepfake technology, and AI will support future campaigns.

South Korea's election regulator allows the use of such "AI candidates" on the condition that it can be identified as a synthesis of deepfake technology and does not spread misinformation.

However, this technology is often restricted.

In 2018, the US news site Buzzfeed and comedian Jordan Peele used deepfake technology to fake a video of former US President Barack Obama, after which Buzzfeed warned netizens not to believe the relevant video information.

South Koreans digitize presidential candidates with deepfake and accept Q&A online

Obama and Jordan Peele.

In 2018, the US news site Buzzfeed and comedian Jordan Peele used deepfake technology to produce a video of former US President Barack Obama, after which Buzzfeed warned netizens not to believe the relevant video information.

Similarly, in 2019, Momo launched an AI face-changing app called ZAO, which rushed to the top of the app list within a few days of listing, but was soon removed due to the potential for abuse. In 2021, a "ants, hey" video generated by face-swapping techniques and its production software Avatarify set off a storm again, and quickly removed the shelves a week later.

In fact, the privacy security issues brought about by face swapping are really worrying. On the one hand, the abuse of technology will lead to the proliferation of false information; on the other hand, if you use the photos of celebrities and others to make relevant videos, it may involve legal issues such as reputation rights, portrait rights, and intellectual property rights. In addition, personal privacy leakage may involve issues such as asset security and personal safety.

Therefore, although deepfake technology has matured day by day, its wide application still needs more security and legal protection.

Reference Links:

https://www.france24.com/en/live-news/20220214-deepfake-democracy-south-korean-candidate-goes-virtual-for-votes

Read on