China AI Safety and Development Association (CnAISDA)

Ce qu'il fait

STATUS DELIBERATELY AMBIGUOUS. Unveiled around the Paris AI Action Summit (Feb 2025) and widely described as China's counterpart to the international AISI network. It is a coordinating network, not an institute: it has no independent staff, no research team and no dedicated budget, and federates existing bodies — Tsinghua University (the intellectual centre), Beijing Academy of Artificial Intelligence (BAAI), CAICT, Shanghai AI Laboratory, the Institute of Automation of the Chinese Academy of Sciences, CCID, and MIIT-linked research groups. Fu Ying described it as 'established with government support'; the precise governmental relationship has never been clarified. Note the Chinese name translates as 'network' (研究网络), not 'association' — the English branding and the Chinese branding do not match.

Ses pouvoirs

convene

Évaluation franche

CnAISDA is real but thin. It exists primarily as a diplomatic interface — something for foreign counterparts to talk to. Its website is English-first, which is itself telling: it is built for foreigners, not for Chinese stakeholders. But I could not extract a single contact address from it. There is no public input process. Practical access for a foreigner is through the individual member institutions or through Concordia AI as intermediary, not through CnAISDA itself.

Notes

It notably calls for 'early warning thresholds for AI systems that may pose catastrophic or existential risks to humans' — the most explicit catastrophic-risk language from any China-linked body. China is NOT a member of the international network; CnAISDA is the parallel structure. Name inconsistency (Association vs Network) is genuine and unresolved.