AI-generated videos mocking Yu Gwan-sun, a revered Korean independence activist, are sparking public outrage as they spread across social media ahead of the March 1st Independence Movement Day.
According to Yonhap News on Wednesday, a TikTok user posted three related videos starting February 22, accumulating over 200,000 combined views.
The videos were created using OpenAI's video generation tool Sora. They depict the independence fighter passing gas, expressing affection toward the Japanese imperial flag, and soaring into space on a "fart rocket." Sora used as reference a photograph taken of Yu in prison clothing during her incarceration at Seodaemun Prison. Her face, swollen from Japanese torture, was restored by AI and used as material for mockery.
Citizens say the content has crossed the line of acceptable humor. Kang, 33, who expressed anger after viewing the videos, said: "I worry that if videos are later made showing the activist saluting the Japanese flag, some people might believe it to be actual history."
Yu Hye-kyung, 61, a great-grandniece of Yu Gwan-sun and head of the Cheonan chapter of the Yu Gwan-sun Memorial Association, said: "It feels like being stabbed in the heart with a knife or awl. Her descendants have lived quietly, careful about every action so as not to overshadow her achievements. This does not seem beneficial for the nation."
In South Korea, AI technology has primarily been used to restore the likenesses of independence activists to promote patriotic awareness. The technology, once praised for bridging the gap between historical figures and modern audiences, has now been repurposed as a tool for disparagement.
Similar controversies have already emerged overseas. In October last year, OpenAI blocked the generation of videos featuring Martin Luther King Jr. on Sora after a fabricated video was created showing him making racist remarks during his 1963 "I Have a Dream" speech.
Historians warn that beyond mocking content, AI's structural characteristics could deepen historical distortion. Because AI chat services deliver information one-way through question-and-answer formats, ordinary users find it difficult to filter out inaccuracies.