Artificial intelligence anthropomorphic interactive services constitute major application scenario of AI application. Such services tend to incur risks， such as emotional dependence， due to their feature of forming emotional bonds with users. These risks can easily trigger security incidents. The newly issued “Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interactive Services” is the first specialized regulatory document in China， which clarifies the features and regulatory objectives of security and serves as the primary basis for guiding and regulating the development of AI anthropomorphic interactive services. However， it exhibits several deficiencies and it’s necessary for us to introduce the objective of “precision regulation” to address those problems. Precision regulation can accord with the refined developmental needs of AI governance and balance technological development and security. A precision regulatory system shall be constructed as follows： first， core concepts should be defined and regulatory boundaries clarified through a combined approach； second， the obligations of direct and indirect service providers should be designed to establish a precise liability regime； third， the factor “mental condition” should be introduced to connect with the civil capacity system under the Civil Code； fourth， a pre-implementation risk assessment mechanism should be established.