智能体存数据黑箱 用户数据流向何处

Group 1 - The year 2025 is referred to as the "Year of Intelligent Agents," marking a paradigm shift in AI development from "I say AI responds" to "I say AI does," with intelligent agents becoming a key commercial anchor and the next generation of human-computer interaction [1] - As intelligent agents approach practical application, the associated risks become more tangible, raising concerns about overreach, boundary violations, and potential loss of control [2] - A survey indicates that nearly 80% of industry professionals are worried about the consequences of user data leakage, highlighting the pressing issue of data transparency and compliance in the era of intelligent agents [2] Group 2 - The transparency of data usage varies significantly across platforms, as illustrated by the example of using AI to generate a resume, where some platforms clearly disclose the tools being used while others do not [3][4] - Different platforms adopt varying approaches to handling sensitive personal information, with some providing clear risk warnings and others obscuring sensitive data in outputs without prior notification [7][9][11] - The responsibility for data handling is often obscured, with user agreements typically placing the onus on developers for the data generated during interactions with intelligent agents [12][13]