[Purpose/significance] The rapid development of Generative Artificial Intelligence (GenAI) products has completely transformed the interaction patterns between humans and artificial intelligence. In-depth exploration of the motives and patterns of self-disclosure in user-GenAI interactions helps to understand the diversity and complexity of user self-disclosure in this context, providing a new perspective for exploring user information behavior in human-intel⁃ligence interaction. [Method/process] Based on the CASA paradigm, social penetration theory, and mind perception theory, systematic thematic analysis was conducted on interview data from 20 GenAI users. [Result/conclusion] The study found that the motives for self-disclosure in user-GenAI interactions can be divided into three dimensions: envi⁃ronmental, technological, and user-related. Moreover, the patterns of self-disclosure in user-GenAI interactions can be categorized into three types: increasing disclosure layer, stable disclosure layer, and decreasing disclosure layer. By ana⁃lyzing the motives and patterns of user self-disclosure, it provides references for optimizing "people-centered" GenAI design, which is helpful for more precise and effective experience optimization for different types of users.