Google shut down 449 YouTube channels in Korean The channels were found to have been involved in "coordinated influence operations," according to a report by the US-based tech giant's threat analysis team. 구글은 8일 자사 블로그를 통해 이 같은 대규모 콘텐츠 삭제 조치를 발표했다. 구글은 “한국어 유튜브 채널 449개가 조직적인 여론조작에 관여한 사실을 발견했다”며 “이를 통해 만들어진 콘텐츠 대부분은 정치적 목적을 띠고 있었다”고 설명했다. 구글은 지난달 29일부터 이달 2일까지 5일간에 걸쳐 449개 채널을 폐쇄했으며, 이 채널들이 올린 동영상 9430여 개도 삭제했다고 밝혔다. 폐쇄된 채널 가운데 38개는 국내 반중 활동가인 이희진씨와 관련된 채널로 확인됐다. 앞서 유튜브는 지난해 6월과 12월에도 각각 12개와 15개의 한국어 채널을 폐쇄한 바 있다. 유튜브 측은 “이번 조치는 기술적 허점을 악용한 조직적 정보 조작 시도에 따른 것”이라며 “앞으로도 이런 유형의 조작 활동에 강력히 대처할 것”이라고 강조했다.

Google has closed down 449 Korean-language YouTube channels in the second quarter of 2026, suspected to be involved in organized online opinion manipulation. According to Google's Threat Analysis Group (TAG) in their "2026 Second Quarter Impact Report," a total of 449 such channels were identified and closed. Monthly breakdowns show that 22 channels were closed in April, 367 in May, and 60 in June, with a significant 81.7% of closures occurring in May alone.

The timing of these closures is raising concerns about an organized online opinion intervention prior to the June 3rd local elections. Google has not directly linked the large-scale channel closures to political schedules. The company has categorized the 449 closed channels into 11 activity networks, indicating that a specific force may have utilized multiple YouTube channels for opinion manipulation activities.

Among these, the largest campaign consisted of 142 channels criticizing the Korean government, which was published in May. Of the closed channels, 255 (56.8%) were critical of the government or political parties, while 173 (38.5%) supported specific political figures. Additionally, 21 channels were found to disseminate both pro-government and anti-government content.

Google defines 'organized online opinion manipulation' as actions where a few individuals use multiple fake accounts to artificially shape public opinion. The company explains that their detection methods focus on technical signals and patterns of behavior rather than the content itself, emphasizing that it is not based on the political views or opinions claimed by the accounts.