China Yanks 14,000 AI Products in "Qinglang" Sweep—Crypto Bots Likely Sweating Too 🧹

China's Cyberspace Administration (CAC) removed more than 14,000 non-compliant AI products, including websites, apps and AI agents, during the opening phase of its 2026 "Qinglang" internet cleanup campaign, the regulator confirmed. The campaign, which began in April 2026, also scrubbed more than 6 million pieces of illegal or harmful information from Chinese networks, suspended over 26,000 accounts and took down more than 1,300 AI-related product listings. Regulators additionally removed nine open-source datasets deemed illegal under Chinese rules. Qinglang, meaning "Clear and Bright," is the CAC's annual internet governance effort, and the 2026 edition marked the first time it targeted AI across the full ecosystem.

The crackdown focused on four main violations: skipping mandatory model registration, weak safety filtering, AI data poisoning and failure to label content as AI-generated across platforms and services. New obligations now require AI services to register, implement safety filters, clearly label AI-generated content and properly manage training data, with non-compliance triggering takedowns and penalties for offending companies.

Major Chinese technology firms moved quickly to comply. Huawei added special reviews in its app store, Alibaba improved its content identification systems, Zhipu built a new review model and DeepSeek added checks to stop data manipulation. ByteDance's Doubao and the Qwen team took a different route, disabling their custom agent features rather than meeting new anti-addiction and instant-exit requirements. Local internet offices also adapted their approach, with Beijing pairing platform self-checks with routine monitoring, Shanghai tailoring rules by platform type, Zhejiang focusing on model auditing and Guangdong building a multi-agency mechanism across the full AI chain.

The CAC said the second phase of the campaign will target AI used to spread disinformation, produce violent material, impersonate people, harm minors and run paid astroturfing campaigns, with heavier penalties promised for offending accounts and institutions. A separate rule, the Interim Measures for AI Anthrop [text cuts off at source], is scheduled to take effect on July 15.