The "Rectifying the Chaos in AI Applications" campaign is a four-month enforcement action launched by the Cyberspace Administration of China (CAC) on April 30, 2026. It targets three categories of conduct: unregistered large-scale models, AI data poisoning, and improper labeling of synthetic content. Rather than introducing new rules, the campaign is an enforcement push against existing requirements under China's generative-AI and synthetic-media regulations (Source: cac.gov.cn).
Status and timeline
The campaign was launched on April 30, 2026 and is scheduled to run for four months, through approximately the end of August 2026 (Source: cac.gov.cn).
Scope
The campaign covers three areas of conduct (Source: cac.gov.cn):
- Unregistered large-scale models — generative AI services operating in mainland China without completing the CAC registration required by the Interim Measures on Generative AI Services (2023).
- AI data poisoning — generative AI services whose training data has been "poisoned" with prohibited content (politically sensitive, culturally inappropriate, or illegal content under PRC law).
- Improper labeling of synthetic content — failure to comply with the content-labeling requirements of the Deep Synthesis Provisions.
Enforcement approach
The action is an enforcement campaign rather than a rule-making instrument: it enforces obligations established by existing regulations, principally the registration regime of the Interim Measures on Generative AI Services and the labeling regime of the Deep Synthesis Provisions. It is described here as an instance of the discretionary-enforcement layer of the "agile governance" approach to Chinese AI regulation discussed in regulating under uncertainty.
Enforcement campaigns of this type have been characterized as tending to impose penalties on a selected set of high-profile cases rather than applying enforcement uniformly, to prompt broader compliance among unselected actors that adjust behavior to avoid being targeted, and to tighten standards in practice without amending the underlying rules through a formal process.
Related actions
The campaign occurred alongside two other PRC actions in April 2026 that bear on AI policy and compute supply. On April 27, 2026, the National Development and Reform Commission (NDRC) blocked the Meta-Manus deal, ordering it unwound. Separately, the Ministry of Commerce imposed export controls on rare earths, a measure adjacent to AI compute supply.
The campaign-style approach contrasts with the single comprehensive instrument of the EU AI Act and with the sector-specific, fragmented approach in the United States.
Relationships
- related: China — Interim Measures for the Management of Generative AI Services (the registration regime the campaign enforces); China — Provisions on the Administration of Deep Synthesis Internet Information Services (the labeling regime); China — Internet Information Service Algorithmic Recommendation Management Provisions (adjacent recommendation-system regulation); Cyberspace Administration of China (CAC); Regulating Under Uncertainty; Synthetic Media / Deepfakes; AI and Content Moderation; Three Theories of Victory (US / EU / China AI Governance).