In the video, a young woman sits in the stands, sighs and turns her head, looking impossibly composed. Some 15 million strangers fall briefly in love – with someone who never existed.
According to the caption accompanying one of many posts, she was “the average Korean woman”. Her admirers quickly crowned her a “baseball goddess”, analysing her every feature with the forensic enthusiasm reserved for internet obsessions, as the five-second clip went viral across South Korea’s online communities.
Then someone noticed the scoreboard. The broadcast graphic in the top left of the picture showed a current Hanwha Eagles pitcher facing a retired Doosan Bears batter in a match-up that could not have occurred.
The goddess was a fiction – indistinguishable, to millions of viewers, from the real thing.
Within days, copycat clips had started flooding social feeds. Users inserted themselves beside BTS star Jungkook in stadium seats and transplanted invented spectators into Formula One paddocks and NBA arenas, all with artificial intelligence (AI).
In South Korea, generative AI has rapidly transformed from a novelty into an everyday tool. Such clips test how quickly synthetic content can pass as reality – and how easily entertainment can slide into misinformation, reputational harm, political manipulation and public confusion.
A wolf that never was
The widening gulf between synthetic image and lived reality was laid bare in April, with the citywide hunt for Neukgu, a two-year-old wolf that had escaped its enclosure at O-World theme park and zoo in central Daejeon.
It triggered a nine-day search. As authorities combed the region, an office worker in his forties used an AI image generator to fake a photograph of a light-brown wolf casually crossing a real intersection near a school.
The image was convincing enough to deceive Daejeon’s own emergency management officials, who incorporated it into public alerts urging residents to stay indoors. It was even displayed at a televised press briefing.
Digital investigators and wire service fact-checkers eventually identified telltale inconsistencies in the image. Police arrested the creator on charges of obstructing official duties. But the damage had already been done: a fabricated wolf had briefly become the face of a genuine public emergency.
“Viewed lightly, this could simply be seen as entertainment or a hobby,” Kim Myuhng-joo, executive director of the Korea Artificial Intelligence Safety Institute, said. “But if this behaviour becomes excessive or prolonged, the gap between fantasy and reality can widen.”
Escaping reality
South Korea has embraced generative AI with lightning speed. Nearly half the population – 44.5 per cent in 2025 – have used such tools, according to a recent government survey.
The penetration rate reflects both the country’s world-class digital infrastructure and a cultural appetite for technological novelty that precedes the current AI moment by decades.
South Koreans also consume more “AI slop” – low-quality, algorithmically optimised content engineered for clicks – than any other population on Earth, according to a 2025 report from video-editing platform Kapwing.
The same generative tools that produce viral baseball goddesses are being used to converse with elderly residents through AI-powered companion dolls, to flood e-book markets with machine-written titles – one US publisher, Luminary Books, reportedly produces up to 9,000 AI-assisted e-books annually – and to churn out “news” content at a pace that human editorial judgment cannot match.
In a country sometimes nicknamed the “plastic surgery republic” for its acute cultural emphasis on appearance and image, AI has found fertile ground as what Kim calls “a tool that provides vicarious satisfaction for suppressed desires and frustrations”: a digital mirror calibrated to reflect something better than the original.
Deepfake epidemic
President Lee Jae Myung has repeatedly sought to position AI as a cornerstone of South Korea’s economic future. In January, the country enacted a comprehensive AI law intended to balance innovation with ethical safeguards across high-impact sectors such as healthcare, transport and finance.
Critics, however, note that its penalties fall well short of those imposed under the European Union’s AI Act: a gap that may prove consequential as the technology’s harms become harder to ignore.
South Korea has already experienced some of those harms in their most visceral form.
In 2024, the country was shaken by a deepfake pornography epidemic, with the mass creation and circulation of sexually explicit AI-generated imagery targeting women and minors, distributed largely through encrypted Telegram chat rooms. Many victims were schoolgirls whose images had been manipulated by classmates.
AI has since become an equally potent political weapon. Fabricated television reports falsely placed a mayoral candidate on Time magazine’s list of rising political leaders. AI-generated K-pop songs were deployed to praise politicians and mock their opponents ahead of elections.
Authorities say they have mobilised hundreds of personnel to monitor manipulated content ahead of next month’s local polls.
The economic consequences are also starting to be felt. Nearly 98 per cent of the country’s 211,000 youth job losses between July 2022 and July 2025 occurred in industries with high exposure to AI automation, according to South Korean news agency AJP, citing Bank of Korea data.
