China Expands AI Safety Enforcement for Open Source AI Communities

China’s cyber regulator, the Cyberspace Administration of China (CAC), has issued a notice tightening enforcement around AI services, specifically targeting the safety management and governance of open-source AI models and developer communities. As part of a 2026 campaign labeled as targeting "AI application chaos," the regulator identified inadequate oversight and safety management of open models as a primary regulatory concern.

The notice criticizes open-source platforms for lacking identity verification, safety controls, and effective review mechanisms, while also calling out failures to promptly remove high-risk datasets or model code. The CAC emphasized that the enforcement action does not constitute a ban on releasing open models, but rather aims to integrate developers into the country’s broader AI safety and registration frameworks. Alongside the notice, the regulator launched a dedicated reporting website to handle whistleblower submissions and track violations ranging from unsafe training data and impersonation to fake engagement operations.