Abstract: 【Purpose/significance】As a key application area of artificial intelligence technology, deep synthesis has increasingly high⁃ lighted the information security risks of its generated content. The existing regulatory system lags behind in terms of governance logic, regulatory tools, and responsibility chain, and there is an urgent need for systematic solutions to fill regulatory gaps【. Method/process】 Through a comprehensive review of existing research results and analysis of the main application scenarios and technical principles, the causes and solutions of the current situation of deep synthesis information supervision are clarified.【Result/conclusion】It is be⁃ lieved that the pre preventive supervision of raw data, the classification and grading of application scenarios, the in-process supervi⁃ sion, and the post supervision system of rights and interests relief should be connected, and the obligation of technology providers to identify content should be strengthened. Agile supervision tools for upgrading classified scenarios should be upgraded, and the legal re⁃ sponsibility chain for artificial intelligence should be improved【. Innovation/limitation】Propose a full cycle deep synthetic information supervision scheme that needs further verification, including generating multi-dimensional spectral labels for information content, pre training key information facility hierarchical governance, and specialized legislation for artificial intelligence.