AI-generated fake doctors flood YouTube: 42 of the 100 most-watched health videos for seniors in Korea

By Seoul Economic Daily · October 21, 2026.

42 of the 100 most-viewed health videos for seniors on YouTube in April 2026 were created with generative AI. Of those, 24 posed as doctors or other healthcare professionals. Only 6 videos in total, 6%, featured real experts. The data come from the Korea Health Promotion Institute, were provided to lawmaker Park Yong-gap of the Democratic Party, a member of the National Assembly's Health and Welfare Committee, and were released on the 20th.

According to the institute, the fabricated videos sought credibility with white coats and specialist titles. Some used the names of nonexistent specialists as channel names, and others staged scenes suggesting the presenter had treated patients in a hospital. The content included medically unfounded claims, such as that "warm water works better than medicine for the stomach" or that "frozen blueberries cure diabetes." Some did not disclose that they were made with AI. According to the institute, AI-generated videos were seven times more likely to contain false information than those by real experts.

The exposed audience is large. According to a survey by the institute itself, 45.4% of people in their sixties look up health information at least once a week. Internet portals are the most common channel (49.7%), and YouTube reaches 14.4%. According to Statistics Korea, 97.2% of seniors spend their leisure time watching programs or videos, and daily use of information and communication devices rose by one hour and five minutes over five years, to 1 hour and 39 minutes.

Following the lawmaker's complaint, Google Korea reviewed 100 videos from 51 channels. It shut down 10 channels for violating its rules, removed 31 videos, labeled 13 that did not disclose AI use and suspended monetization of 10 videos from five channels. Since July, YouTube has also limited monetization of low-quality, mass-produced AI videos and sensationalist content. That policy includes AI avatars based on real people giving medical, financial or legal advice.

Public oversight has gaps. The institute can recommend voluntary corrections to producers or platforms, but according to the data it has never requested that a video be deleted. The lawmaker proposes a coordinating body of public institutions, platforms and medical experts for regular inspections and proactive blocking of harmful videos.

**Our take.** This case shows a short-term AI risk that requires no superintelligence: producing false trust at almost zero cost. A white coat and a calm voice used to be a reasonable signal of authority, and now they can be generated in minutes. If face and voice no longer serve as proof, verification has to move elsewhere: the sender's accredited identity, content traceability and mandatory labeling.

The platform's response is revealing. According to Seoul Economic Daily's report, YouTube's measures mainly target advertising revenue, which is why continuous oversight is being called for to curb the circulation of falsehoods itself. Removing the financial incentive helps, but it doesn't stop a video with false advice from reaching a senior with diabetes who believes it and delays seeing a doctor. A caveat is also in order: the figure comes from a sample of 100 videos from a single month, and we don't know how each one was classified as AI-generated, so it serves as a warning sign rather than a definitive measurement.

There will be losers and winners. The losers are seniors and, with them, the health systems that will end up treating cases that arrive too late. The winners are industrial content producers who exploit attention. And those offering credible verification will also win, such as medical associations, accreditation seals and platforms that link every piece of clinical advice to a verifiable professional.

In general, the same technology can be part of the solution in the long term: medical assistants with verifiable sources and clinical oversight, bringing reliable information to those who today lack easy access to a doctor. That horizon, of better and more accessible health thanks to AI, can only be reached if trust in health information is not eroded first. The transition demands identity and labeling rules that don't wait until the damage is already done.