China's internet regulator removed more than 14,000 AI products in the opening phase of its Qinglang campaign, according to reports from The Economic Times. The Cyberspace Administration of China (CAC) launched the annual internet governance campaign in April 2026, marking the first time AI products were targeted across the entire ecosystem. The campaign, named 'Qinglang' meaning 'Clear and Bright', focused on removing harmful or illegal AI content and implementing stricter compliance measures. The regulator has now set up a dedicated reporting platform to receive public complaints regarding AI-related abuses, demonstrating enhanced public participation in the enforcement process.
The first phase of Qinglang resulted in significant removals across multiple AI categories. As reported by The Economic Times, the CAC suspended over 26,000 accounts and took down more than 1,300 AI-related product listings. The regulator also removed nine open-source datasets deemed illegal under current Chinese rules. Additionally, the campaign scrubbed more than 6 million pieces of illegal or harmful information across Chinese networks, demonstrating the comprehensive nature of the cleanup operation. Local internet regulatory authorities have implemented targeted measures based on regional conditions, with Beijing establishing a coordinated mechanism combining platform self-inspection, routine monitoring, technical screening and case verification, leading to swift removal of non-compliant AI applications and accounts.
According to Reuters, Beijing may be considering restrictions on overseas access to China's most advanced AI models, including future releases. Alibaba, ByteDance and Z.ai were reportedly among companies involved in recent discussions with Chinese authorities regarding these potential controls. The move would represent a significant shift from China's open-model ecosystem approach, treating frontier models less like commercial software and more like strategic infrastructure. This could provide a modest reprieve for the US AI complex as Chinese models' price-disruption threat becomes less immediate for US frontier-model providers, though it would not eliminate broader token-pricing pressures or capability convergence concerns.
The campaign targets four main problem areas including skipping mandatory model registration, weak safety filtering, AI data poisoning, and content not properly labeled as AI-generated. According to reports from The Economic Times, new obligations now apply across the board requiring AI services to register, implement safety filters, clearly label AI-generated content, and properly manage training data. Failure to comply can trigger real takedowns and penalties for offending companies, marking a significant shift in AI regulatory enforcement. The CAC has announced that the second phase will focus on curbing the misuse of AI to produce and disseminate disinformation, spread violent and vulgar content, impersonate others, infringe upon minors' rights, and engage in online astroturfing activities.
Major Chinese technology firms have moved quickly to comply with the new regulations across different regions. As reported by The Economic Times, Huawei has added special reviews for generative AI service registrations and labeling in its app store, while Alibaba has improved digital fingerprint comparison and keyword interception mechanisms to enhance content identification accuracy. Zhipu has developed a multimodal auditing model for complex scenarios, and DeepSeek has embedded anomaly detection algorithms in data collection and preprocessing to prevent manipulation. Shanghai has refined regulatory approach by platform type and upgraded reporting mechanisms, while Zhejiang Province has carried out targeted rectifications on model auditing capabilities, training data security and AI data poisoning prevention. Jiangsu Province has opened a special reporting channel covering five categories of violations, and Guangdong Province has established a multi-department coordination mechanism for full-chain governance.
