Recently, several industry insiders have indicated that major social media platforms like Weibo and Tencent have significantly strengthened their content review process. Due to the increase in automated deletion triggered by machine recognition, cases of mistaken deletion have become more apparent. According to a platform administrator, in the past three months, there has been a 20% increase in mistakenly deleted posts due to machine recognition errors, including mistakenly deleted and banned official posts.
Multiple individuals familiar with the platform’s content review mechanism have told reporters that in the recent period, the proportion of erroneous deletions caused by machine misjudgments has noticeably risen. The number of mistakenly deleted posts has increased by approximately 20% compared to the previous period. In addition to regular users, some official accounts have also experienced content deletion or bans.
Chen Yang, an industry insider who has been in regular contact with platform review mechanisms, told reporters on December 15th: “After receiving complaints, administrators found through manual checks that some of the deleted content did not involve clear violations but were deemed risky by the system based on keywords, repost frequency, or dissemination paths.”
Chen Yang further stated to reporters: “In November, our platform (the interviewee declined to be named) received around two to three hundred thousand complaints about mistakenly deleted content, mainly due to the addition of keywords and sensitive words that the administrators couldn’t handle in time.” He added that the machine review had been “overzealous,” leading to a noticeable increase in misjudgments.
Recently, an incident involving the suspension of an anti-fraud account by the public security system in Yulin City, Shaanxi Province, has attracted attention. Screenshots circulating online show that the Yulin Anti-Fraud Center reported that their WeChat account used to disseminate anti-fraud information to community groups and residential areas was flagged as violating regulations and subsequently banned by the system. The article mentioned that the account primarily shared anti-fraud guides, case studies, and risk warnings and was not a regular social media account, but multiple appeals did not receive a clear explanation.
The screenshot displayed an article titled “To Tencent Company: Our Anti-Fraud Propaganda Account…” signed by Yulin Anti-Fraud Center, with a publication date of December 5, 2025. The article indicated that the account was considered an essential tool for grassroots anti-fraud alert, and after the ban, the dissemination of anti-fraud information was temporarily affected.
Subsequently, the platform’s customer service responded that after manual review, the account exhibited “irregular behavior.” However, the specific violation content or terms breached were not disclosed in the response, and the reason for the ban remains unclear.
Luyao, a mainland operator of self-media, told reporters that recently, cases of mistaken deletions are not limited to the Yulin incident: “One of my friends posted some screenshots about anti-fraud on Weibo, including an anchor talking about anti-fraud and online fraud, and they got removed.”
Luyao revealed that posts from some state media client platforms had also been deleted by the system, but were later resolved through internal channels and communication with the platforms.
Another industry insider pointed out that mistaken deletion cases are more concentrated at the local level, with local government departments and public security system accounts being the more affected groups. Netizen Mr. Lu said: “I have also heard that some local public security bureau accounts’ contents were frequently deemed a violation, and they could only appeal through official channels or find acquaintances to negotiate with platforms like Weibo and Tencent before being unblocked. Now even the police are hesitant to comment casually.”
The tightening of social media platform review measures in mainland China is also affecting users’ posting behavior. Mr. Lu, who has been using social media platforms for a long time, frankly stated that due to the fear of content being incorrectly judged by the system: “Many people are no longer easily posting text posts and are opting for screenshots, pictures, or external news links to reduce the risk of deletion.”
Multiple interviewees have noted that the current platform review mechanism heavily relies on machine recognition and, with limited human review resources, the cost of misjudgments has been shifted to content publishers, leading to a tightening of posting behavior. As of now, mainland social media platforms have not provided any explanation regarding the recent changes in mistaken deletion rates or adjustments to the review mechanism, and there have been no reports from the media on this matter.