China NDRC says AI lawmaking will accelerate as domestic large models top 10 billion downloads

China’s National Development and Reform Commission said the country will speed up work on AI legislation while expanding research and pilot infrastructure for the sector. Speaking at a July 31 press conference, Jiang Yi, director of the Policy Research Office at the NDRC, said China’s homegrown AI innovation gained pace in the first half of the year. He cited local companies including DeepSeek and Moonshot AI, saying they had released open-source large models with parameter counts at the trillion level. He also said cumulative global downloads of Chinese large models had exceeded 10 billion. Jiang said the next step will focus on frontier areas including foundational theory for large models, training and inference efficiency, multimodal capabilities, and AI agents. Authorities will also move faster on the layout and construction of national AI application pilot testing bases and step up efforts to foster AI-derived businesses. On regulation, he said policymakers will coordinate high-quality development with high-level security, accelerate the legislative process for an Artificial Intelligence Law, build a closed-loop framework covering development, regulation, and safety, and improve mechanisms for technical monitoring, risk alerts, and emergency response. He added that China will adhere to a development approach in which AI is human-led, used for people, and aligned with human well-being, while addressing its impact on economic operations and employment distribution.