2024-5, the promulgation and implementation of the new Law on Guarding State Secrets had a major impact on the management of classified archives in China. Although the open review of archives in China has been carried out for a long time and has accumulated considerable practical experience, the declassification of classified archives in China is still at a low level. In response, this project project focuses on new-generation artificial intelligence, analyzes the feasibility of its application in the declassification and downgrading of archives, and provides an application demonstration. First, it systematically reviews the current status of archival declassification in China and identifies the key problems, to provide theoretical guidance for the research of this project. Next, to address problems in declassification work such as semantic ambiguity, inconsistent standards and poor adaptability, it carries out research on a standard system that supports structured modeling. Then, building on the research foundation of previous projects and the results of earlier surveys, it creates a new-generation AI archival declassification and downgrading technology system. Finally, it delivers an AI-assisted tool for the declassification and downgrading of archives that meets the actual needs of the archives, providing an exemplary reference for application in other archival institutions.