China's New Multichannel Content Rules Take Effect as Regulator Reports Removing 5.6 Million AI Posts

China's internet just got a lot more regulated, and it happened in two moves within days of each other.

On Sept. 1, 2026, new rules from the Cyberspace Administration of China (CAC) and four other government agencies took effect, formally titled the Provisions on the Management of Multichannel Distribution Services for Internet Information Content. Companies that produce, distribute, market or manage online talent now have to register as licensed service providers, according to The Epoch Times. Firms already operating have 30 days to amend their business registrations.

Two days later, on Sept. 2, the CAC announced results from a separate but related campaign: 5.61 million pieces of AI-generated or AI-related content removed, more than 49,000 accounts penalized, and over 2,400 websites and apps shut down, according to China Daily and TechRepublic. That four-month effort, part of the CAC's ongoing "Qinglang" campaign launched in April, targeted AI content used to spread false information, impersonate real people, or endanger minors, TechRepublic reported.

Under the Sept. 1 provisions, companies must appoint a designated content manager, build a management team, verify the identity of every creator they contract with, and spell out who's responsible for what content. Platforms can restrict account functions, cut off monetization, or shut accounts down entirely if a company or creator is found violating "laws, regulations, platform rules, or contractual agreements," per The Epoch Times.

The rules also set hard age limits. Minors under 16 are barred from livestreaming entirely, and those 16 to 18 need parental or guardian consent plus age verification before they can go live, according to netinfluencer, which cited the CAC's list of 11 specifically banned activities meant to stop "rumors" and "cyberbullying."

An independent content creator in Shenyang told The Epoch Times the vagueness is the real problem. "Many people doing livestreams don't know what they can say and what they can't say. Once the traffic gets high, the risks come with it," she said. She said she's watched other creators get their accounts blocked once their livestreams started drawing real audiences, and she questioned who decides what counts as "inciting emotions" or "false propaganda." "There is no clear standard," she said.

Other creators and critics who spoke to The Epoch Times, on condition of anonymity out of fear of reprisal, said the rules could be used to suppress reporting on unemployment, unpaid wages, unfinished housing projects, and protests, not just to fight actual fraud.

Netinfluencer, citing the South China Morning Post, reported that Chinese cyber police recently punished two internet users for posting derogatory comments mocking a village official over a dispute involving gold earrings. That's the kind of enforcement action critics point to when they say "anti-rumor" rules end up policing criticism of local officials, not just scams.

China Daily, the state-run outlet, frames the AI crackdown almost entirely around protecting vulnerable people. It cites accounts penalized for using AI to fake disaster and rescue footage, deploy digital avatars to swindle elderly users, and use face-swapping or voice-cloning to impersonate public figures or produce sexually suggestive content for profit. Those are real harms, and stopping AI-generated deepfake scams targeting seniors is a legitimate goal that deserves consideration on its own terms.

But China Daily's write-up doesn't mention creator concerns, doesn't quote a single critic, and doesn't touch the broader multichannel licensing regime that took effect the same week. TechRepublic filled in more of the mechanics, reporting that platforms including Douyin, Kuaishou, Weibo, Tencent, Baidu and Xiaohongshu have expanded facial and voiceprint detection libraries and issued 46 governance notices, and that chatbot makers ByteDance's Doubao, Tencent's Yuanbao, Alibaba's Qwen and Baidu's Ernie Bot are now restricting what their models can output at the training-data stage, not just after publication.

The practical effect for anyone building an audience in China is that the line between "stopping scams" and "managing what people are allowed to say" runs through the same set of rules, enforced by the same regulator, with penalties decided by platforms and cyber police rather than courts. Whether that line holds where the CAC says it does is the question independent creators are now living with, one livestream at a time.