Naver News Comments Used for Public Opinion Manipulation Aimed at 'Causing Conflict' Beyond 'Factional Logic'

Comment work is considered a job rather than a mere hobby, and there is a set price for it. The host emphasized, “If you piled up posts in the same spot every day, that’s not a hobby, it’s work, and work comes with a price.” Cases of public opinion manipulation originating from China have confirmed the spread of specific agendas, such as data center electricity issues, and a structure has been identified where private companies contracted by Chinese local governments act as proxies for such manipulation.

According to an investigation by Chinese state television, 10,000 comments are traded for 1,680 yuan, which is approximately 350,000 Korean won. This sets the market price at about 35 won per comment line. This public opinion manipulation operates in a structure where not only official companies but also individual operators profit.

In a case of an illegal viral organization busted by the Busan Metropolitan Police Agency last July, approximately 29,000 fake reviews were identified, and their criminal profits amounted to a staggering 1.9 billion won. Such organizations have a four-stage division of labor system: advertiser, agency, developer, and executor, with the actual reviewers receiving about 700 to 800 won per review.

According to Harvard University research, China's 'Wumao Party' (50 Cent Party) consists not of part-timers paid per post, but of government employees on salary. They execute public opinion manipulation through a dual structure where external private companies and internal government officials operate simultaneously. The purpose of public opinion manipulation is not to persuade the opponent, but to guide public opinion in a specific direction and simply carry out commissioned tasks.

In the KAIST study, 7 of the top 10 suspicious accounts' targets were Korean politicians, targeting both progressive and conservative sides indiscriminately. The host explained, “If public opinion manipulation were about taking sides, this list should be skewed to one side. But in reality, far from being skewed, it is evenly spread across both camps.” This suggests that the ultimate goal of public opinion manipulation is to cause social conflict and division, rather than to support a specific candidate.

An analysis revealed that accounts newly created or activated during major election periods were 51% more numerous than usual, and the paper judged this to be a deliberate schedule, unlikely to be a coincidence. The host noted, “Accounts newly activated during major election periods are 51% more numerous than usual. The paper explicitly stated that this difference is too significant to be a coincidence.” There is also a tendency for activity to commence during times of heightened social conflict, such as the Lunar New Year holiday.

The purpose of manipulated posts is not to win an argument but to divert attention and shift people's focus elsewhere. A strategy is used to push up completely different, sensational topics rather than refuting, thereby blocking existing attention. The host explained, “These posts weren’t written to win an argument, but to make people look away from what they were currently focusing on.” Ultimately, the goal of public opinion manipulation is to make us distrust others first.

OpenAI, in a report on June 10, 2026, disclosed two fake news operations that utilized ChatGPT. These operations generated numerous posts related to electricity prices and US-China trade. It was revealed that these posts were written by ChatGPT, not humans, and published by deceptive accounts posing as Americans. Traces of instructions entered in simplified Chinese by the deceptive accounts were detected, and specific instructional sentences dealing with Trump and Xi Jinping were also found.

According to OpenAI's classification criteria, these operations remained at spread levels 1 and 2, and there was no evidence that the targeted individuals reacted significantly. Ben Nimmo, who led the investigation, emphasized that these activities were more experimental in nature than influential. The direct exposure of instructions was attributed to the use of external services.

According to internal documents leaked from a company under the Chinese Academy of Sciences (reported by the New York Times), this company uses publicly available AI models like Deepseek to create fake personas whose speech patterns change according to the listener. This is an automated system that databases personal information of US congressmen, journalists, and scholars, and adjusts speech patterns accordingly. It was explained that "what is said to a congressman and what is said to a journalist comes out differently from the start." Such technology is evolving public opinion manipulation by reducing costs and leaving no traces.

The revised espionage act passed in February of this year expanded its scope to include foreign entities and equivalent organizations, and an amendment to the Act on Promotion of Information and Communications Network Utilization and Information Protection now allows for 5x damages for the dissemination of false or manipulated information. However, the host pointed out, “This means that both new laws were not created to withstand this kind of activity. This is as far as our current measures go.” The current laws focus on espionage and large-scale accounts, making them insufficient to block general comment manipulation.

Naver applied Cleanbot to all news articles, including political and election news, starting this spring. If malicious comments exceed a certain threshold, the comment section for that article is completely closed. However, as criticized, “What gets hurt is the comment section, not the account,” information that could allow tracking, such as account nationality or access logs, is still not disclosed. This remains a blind spot, making it difficult to identify the perpetrators of public opinion manipulation in Korea, unlike with overseas companies like Google and Meta.

Last month, the Hitotsubashi University Institute in Japan uncovered that Chinese-affiliated fake news sites were using Korean news articles as material to sow division in Japanese society. This once again shows that the purpose of comment manipulation goes beyond simple public opinion manipulation to inducing mutual distrust and conflict. Such fake news, spreading across borders, risks escalating into international conflict as well as within Korean society.