China Removes 14,000 AI Products in Qinglang Regulatory Sweep
The Cyberspace Administration of China's first AI-focused cleanup campaign suspended accounts, removed datasets, and imposed new compliance requirements on tech giants.
China Expands Internet Governance to AI Systems
China's Cyberspace Administration (CAC) removed more than 14,000 AI products from domestic networks during the initial phase of its 2026 Qinglang campaign, according to details first reported by BeInCrypto. The annual "Clear and Bright" internet governance initiative targeted AI systems across the country's technology ecosystem for the first time, marking a significant expansion of regulatory oversight.
The enforcement action extended beyond product removals. The CAC scrubbed over 6 million pieces of content it classified as illegal or harmful, suspended more than 26,000 accounts, removed over 1,300 AI product listings, and deleted nine open-source datasets deemed non-compliant with Chinese regulations.
Four Core Violations Triggered Enforcement
The campaign, which launched in April 2026, focused on specific compliance failures. Regulators targeted AI services that skipped mandatory model registration, implemented inadequate safety filtering, allowed AI data poisoning, and failed to properly label AI-generated content.
New requirements now apply uniformly across the sector. AI services must complete registration processes, deploy safety filters, clearly mark AI-generated content, and demonstrate proper training data management. Non-compliance now carries enforcement consequences including takedowns and penalties.
Tech Giants Implement Compliance Measures
Major Chinese technology companies responded with immediate compliance efforts. Huawei introduced specialized review processes in its app store, while Alibaba enhanced content identification systems. Zhipu developed a new review model, and DeepSeek added safeguards against data manipulation.
ByteDance's Doubao and the Qwen team took a different approach, disabling custom agent features rather than implementing new anti-addiction and instant-exit requirements. Regional internet offices adopted varied enforcement strategies, with Beijing combining platform self-audits with routine monitoring, Shanghai tailoring rules by platform category, Zhejiang emphasizing model auditing, and Guangdong establishing multi-agency coordination.
Second Phase Targets Disinformation and Minors
The next enforcement phase will address AI systems used to spread disinformation, generate violent content, impersonate individuals, harm minors, and conduct paid astroturfing operations. The CAC indicated it will impose stricter penalties on violating accounts and institutions.
Separate regulations taking effect July 15 will govern AI anthropomorphic interactive services designed for emotional relationships. The Interim Measures for AI Anthropomorphic Interactive Services prohibit virtual-companion services for minors and mandate guardian consent for users under 14.
Why it matters
This regulatory expansion demonstrates how China is applying its established internet governance framework to emerging AI technologies at scale. The enforcement approach—combining mandatory registration, content filtering, and age restrictions—creates a compliance template that could influence AI regulation in other markets while potentially fragmenting the global AI development landscape. The timing coincides with intensifying US-China AI competition, where Chinese firms have demonstrated the ability to match new American systems within months of release.
Details of the Qinglang campaign and enforcement actions were first reported by Luis Blanco at BeInCrypto.
This is an original analysis by the Omega editorial team. Source reporting: AI Watch.
Want systems like this working for your business?
Book a Call
