China's AI clean-up campaign reports a second round of results
Enforcement is running alongside the rulebook, with action published in July and September.
Background
China's Cyberspace Administration launched a four-month 2026 AI application rectification campaign targeting failures in content labelling, filing, safety review and data practices. Two rounds of results were published: in July and September 2026.
Three key points
1. What the campaign targeted. The campaign took action against AI products, user accounts, platforms and other services for failures including non-completion of required filing, missing safety reviews, data poisoning, and content labelling violations. Enforcement was directed at both the technical and operational layers of AI services.
2. Two rounds of published results. First stage results were published on 6 July 2026 and second stage results on 2 September. Publishing named results is itself a regulatory tool in China: named enforcement carries reputational consequences alongside any formal penalty.
3. Enforcement runs alongside rulemaking. China has been expanding its AI rules quickly, through generative AI measures, mandatory labelling rules and anthropomorphic AI regulations. The 2026 campaign shows that enforcement is active during the rulemaking period, not waiting for a completed framework.
Why this matters
Brands and agencies that use AI to produce or distribute content in China face an active compliance environment, not just one in waiting. The combination of written rules and named enforcement actions means the risk of non-compliance is real and visible. Labelling and filing obligations are the place to start.
Sources: GAEEDU, China AI governance profile.
