The Chinese regime’s Cyberspace Administration recently announced the closure of nearly 100,000 “self-media” accounts, marking a new round of crackdown on self-media accounts and further squeezing the space for public expression.
According to a notice from the Central Cyberspace Affairs Commission of the Chinese Communist Party on May 3rd, some “self-media” accounts failed to properly label the sources when publishing politically sensitive and public policy-related information, or did not indicate that the content was generated by artificial intelligence. The Cyberspace Administration has directed platforms to deal with over 98,000 accounts. The notice specifically mentioned accounts on platforms such as Kuaishou like “Farmer Sister” and “Tsinghua Fish Mom,” as well as accounts on Bilibili like “Business Strategy,” accusing them of not providing proper attribution when interpreting policy information related to agriculture, education, and elderly care, potentially leading to misunderstandings during dissemination.
A self-media blogger from Dalian, who goes by the pseudonym Xiao Xiao, told Epoch Times that several of her accounts were permanently shut down recently. She said, “Just before May 1st, several of my accounts were banned without prior notice. This time, the crackdown is ostensibly targeting issues with labeling information, but the real change lies in the tightening of rules on information dissemination. Many public issues already lack transparent sources, and once ‘must be labeled’ is required, unofficial narrative spaces will be further compressed.”
Xiao Xiao stated that she only posted videos online that she saw in chat groups discussing topics like falling housing prices, the challenges faced by food delivery drivers, and someone saying they were hungry. She added, “Now I’ve found out that even these kinds of distressing videos cannot be shared.”
The official notice also mentioned another category of accounts, including Weibo’s “Golden Retriever Healing Series” and Bilibili’s “Oh-Oh Little Ranzi,” which use artificial intelligence technology to create videos depicting scenarios like “Golden Retriever Holding a Child” and “Gorilla Confronting a Crocodile,” but fail to indicate that they are AI-generated. The Chinese authorities believe that such content can make it difficult for some netizens to differentiate between virtual and real scenarios while consuming the content.
A prominent internet user and micro-business owner, Liu Lu, stated that this so-called crackdown is aimed at harvesting dissenting accounts and alleged “spy” accounts on a large scale. He said, “Now, AI has become the trend, but the ‘unidentified sources’ and ‘unclear labeling’ thrown out by the authorities are nothing but two invisible nooses around the necks of creators. The boundaries of these concepts are like rubber bands, in their hands, with all the flexibility; they say you violate the rules, so you do violate the rules, they say you spread rumors, so you spread rumors. Now there’s even the suspicion of espionage reporting, it’s too sensational, the Chinese Communist Party wants to make all discordant voices disappear.”
From the initial crackdown on “rumors” to later targeting “Big Vs” (influential internet users) and now focusing on “self-media” for specific clean-up actions, the regulatory justifications have continuously evolved. The so-called “unidentified sources,” “lack of labeling,” or “fabricated deductions” often morph in practice into the highly centralized “digital pocket crimes” of interpretation.
One netizen commented in a chat group, “It’s not about the authenticity of the content now, it’s about what you can or cannot say, it’s about whether you love your country or not.” Another person wrote, “(The Chinese authorities’) cleanup is just fear, it’s about being afraid of people thinking independently.”
Regarding the Chinese regime’s crackdown centered on “AI labeling,” veteran media professional Xia Qiang, speaking under a pseudonym, said in an interview that while AI labeling may help in distinguishing false information in form, in practical execution, it may be used for content management. He remarked, “Technical issues and content management are being merged by the authorities, and the result is the further extension of regulatory boundaries.”
In recent years, the Chinese regime has been intensifying its control over online content, from real-name registration for accounts, algorithm recommendations, to short video content reviews, expanding the scope of supervision. With the increasing popularity of artificial intelligence-generated content, the authorities have incorporated “AI labeling” into their regulatory focus.