China's CAC Pulls Over 14,000 AI Products in Crackdown

The Cyberspace Administration of China (CAC) launched a nationwide, four-month campaign on April 30, 2026 targeting misconduct in AI applications, and reports say more than 14,000 AI products and services have since been removed. The campaign is part of the recurring "Qinglang" (Clear and Bright) special action and runs in two phases. According to initial reporting, the first phase targeted incomplete mandatory filings for large language models (LLMs), deficient safety and filtering systems, improper management of training data, inadequate labeling of AI-generated content, AI data poisoning, security risks in open-source AI communities, and misuse of AI agents and "digital humans." The second phase focused on harmful AI-generated content and online behavior such as disinformation, impersonation, spoofing of historical figures, and sexual or bullying content, removing illegal material and penalizing non-compliant accounts and platforms.