AI-generated fake news leave Koreans confused, angered

A screengrab from a YouTube channel featuring artificial intelligence-generated medical information. A viewer comment says the speaker’s hand gestures look robotic. Captured from YouTube

“AI-generated images have become so realistic that it is nearly impossible to tell them apart,” Yoon, 33, said with a deeply concerned look.

His parents recently showed him a YouTube video, saying they wanted to buy supplement pills promoted by a doctor in the clip. The doctor’s speech and facial expressions appeared so natural that even Yoon almost fell for it, only realizing at the last moment that the video had been generated by artificial intelligence.

“A quick search was enough to show that the advertised pills had no scientific basis, but I worry my parents may fall for AI content again,” he said.

Yoon’s case is one example of the growing dark side of AI technology, as the boundary between reality and fake content becomes increasingly blurred. AI-generated content is beginning to pose a real risk of deceiving people by featuring figures presented as professionals or mimicking television news broadcasts.

Reports of financial and other harm caused by deceptive ads and fake news that conceal their use of AI-generated images continue to emerge, fueling public concern and calls for stronger safeguards.

A restaurant's promotional post staged with artificial intelligence to make it appear as though the establishment had been featured on a television program / Online capture

Patrons of businesses that rely heavily on customer reviews, such as hair salons and plastic surgery clinics, were among the first to encounter AI. Artificially generated fake before-and-after images caused anger and controversy across online spaces.

Kim, a 24-year-old college student, said she grew skeptical after visiting a hair salon’s website and coming across what appeared to be an AI-generated model.

“Hairstyles are hard to judge when AI is involved because people often rely on first impressions,” she said. “Using fake images makes it seem like they are trying to deceive consumers, and it really damaged my trust in the beauty industry.”

Similar suspicions have surfaced in online plastic surgery communities, where users pointed out that after photos for rhinoplasty procedures all appeared to look the same.

Restaurants have also faced strong backlash for uploading AI-generated images that misled consumers into believing they had been featured on television news and cultural programs. One image caused confusion because a broadcaster’s logo was clearly visible on the screen, but the woman appearing to praise the food as “clean-tasting and well-presented” was fake.

“Such acts of impersonating news media could constitute criminal offenses and lead to liability for property rights infringement, defamation and the spread of false information under the Information and Communications Network Act,” said Chung Sang-jo, professor emeritus at Seoul National University School of Law.

A notice posted at a community service center in Eunpyeong District, Seoul, last month warns residents about the spread of false information. Korea Times photo by Na Min-seo

The fallout from AI-generated fake news has even reached government and administrative institutions, where front-line civil servants say they are being worn down by a flood of complaints and inquiries.

“So many residents come in asking why they are not eligible for support programs they saw on YouTube,” said Na, who works at a complaints counter at a community service center in Seoul’s Eunpyeong District. “Having to explain each case one by one has added significantly to our workload.”

Residents, too, are bearing the cost. An 81-year-old recipient of basic livelihood benefits said he visited his local community service center after watching a YouTube video claiming he would receive millions of won in disaster relief payments.

“But I was told I could only receive 550,000 won. I don’t know what is true,” he said.

Fake images also hampered the operation to capture a wolf that escaped from Daejeon O-World Zoo last month, after they were submitted as supposed tip-off material. Some 503 personnel and 61 vehicles were dispatched to an unrelated location, causing officials to miss a critical window to catch the wolf and prolonging the operation.

A Daejeon Fire Headquarters official said authorities had no way to immediately determine whether the submitted images were AI-generated, and that zoo staff at the scene had identified the animal in the images as the escaped wolf. A man in his 40s who allegedly created and distributed the images online was taken into custody on suspicion of obstruction of official duties.

An AI-generated image appears to show the wolf that escaped from Daejeon O-World roaming through the city. Courtesy of Daejeon Fire Headquarters

The series of incidents has prompted the government to take steps to address AI-generated fake news.

The Basic Act on Artificial Intelligence, which requires AI service providers to label generative AI output and deepfake content, took effect in January. In February, the government also held an inter-agency ministerial meeting on responding to fake news and other abuses of AI, announcing plans to develop deepfake detection technology and expand media literacy education.

But experts say that as the use of AI expands rapidly, users’ own efforts to verify information and act responsibly are just as important as detailed laws and systems.

"AI detection technology needs to be advanced in pace with AI technology, and more specific response measures should be prepared within each sector," said Oh Jung-ik, a lawyer at One Law Partners.

Choi Byung-ho, a research professor at Korea University’s Human-inspired AI Research Institute, said the specific standards for applying the law should be refined through discussions among relevant stakeholders.

“We are now living in an era where a healthy skepticism toward whether something is real can actually benefit individuals,” he said.

This article from the Hankook Ilbo, the sister publication of The Korea Times, is translated by a generative AI system and edited by The Korea Times.