Abstract: The uncertainty of artificial intelligence not only serves as the consensus-driven motivation behind the global wave of artificial intelligence legislation but also acts as the divergent factor that triggers the multifarious legislative practices across nations.The European Union,the United States,and Japan have respectively formed three distinct legislative models:risk regulation emphasizing government supervision,industry self-governance emphasizing market dominance,and soft law guidance focusing on government-enterprise cooperation.However,all these three legislative models view law as a tool for the government to enforce specific governance concepts,thereby generating an overconfidence that either the government or the market can independently handle the uncertainties of artificial intelligence.For China’s unified legislation on artificial intelligence to overcome the predicament where a single entity finds it difficult to regulate uncertainties and to achieve the dynamic governance goal based on communication among multiple subjects,it is necessary to rely on democratic communication based on risk symmetry rather than planned regulation to control uncertainties.In this regard,the unified legislation on artificial intelligence can be based on the institutional design of communication procedures to urge the government to eliminate risk asymmetry in the market,ensuring that multiple stakeholders in specific scenarios make agile and accountable risk decisions through risk communication under conditions of risk symmetry,thereby achieving dynamic governance of risk sharing.