DistantNews
Support us

China’s ‘account attacks’ turn AI content moderation loopholes into a business

From Hankyoreh · () Korean

Translated from Korean and summarized by DistantNews. Read the original for the full story.

At a glance

News Named sources Ongoing story
  • Operators on Chinese short-video platforms such as TikTok, Douyin and Kuaishou are reportedly turning mass reporting and account blocking into a black market called “account attacks.”
  • They allegedly threaten creators with further reports, demand payment, or remove unfavorable content for businesses and brands.
  • Automated moderation often acts quickly on fabricated privacy or copyright complaints, while appeals from targeted users can take days and may be rejected.

On China’s short-video platforms, getting a rival’s video removed or account blocked has become a service that can be bought and sold.

The practice, known as “dahao,” or “account attacks,” exploits the limits of automated content moderation. Platforms including TikTok, Douyin and Kuaishou use artificial intelligence to identify rule-breaking content quickly, but the systems do not always verify whether reports are genuine or understand their context.

Operators concentrate reports on a creator’s videos, trigger removals or account suspensions, and then demand money. They also accept payments from companies and brands seeking to eliminate content that harms their interests. WORLD, an art creator with 260,000 followers, said one account was blocked in June and videos on another account were then reported. The reporter threatened further action unless WORLD paid. After WORLD sent 1,000 yuan, the report was withdrawn and the video restored.

Platforms tend to remove videos first in the initial AI review because of pressure to comply with regulations, and then review them again.

· Security expertThe expert explained why automated moderation benefits malicious reporters.

Online advertisements offer 500 reports for 25 yuan and guarantee a video’s removal for 1,200 yuan. One operator boasted of earning 53,715 yuan in a single month. Operators reportedly take orders through overseas social media and recruit others, teaching them platform-specific reporting methods and wording.

The system’s imbalance favors the attackers. Operators fabricate claims of privacy or portrait-rights violations and submit reports generated with artificial intelligence. In one test, a video with no actual privacy violation was reported as a privacy leak and removed within two hours without firm evidence. Victims may submit identification and supporting documents, wait days, and still lose their appeals, while the complainant can restore the video simply by withdrawing the report. A security expert said platforms often remove content first to meet compliance pressure, leaving fast reports and slow appeals as an advantage for malicious complainants.

The imbalance in cost between fast reporting and slow appeals creates an environment favorable to malicious reporters.

· Security expertThe expert described the structural weakness in the appeals process.
About this summary

Originally published by Hankyoreh in Korean. Translated, summarized, and contextualized automatically by DistantNews, with a note on how the source frames the story. Not individually reviewed before publishing. How this works.