以ChatGPT为代表的大语言模型促进了人工智能新范式的形成，同时也引发了人们对以人工智能为中心的信息处理的恐惧和担忧，社会公众对人工智能引发的伦理问题呼声日益强烈。文章首先梳理了人工智能伦理相关文献，同时对全球官方发布的115项人工智能准则进行了整合性综述，剖析了人工智能伦理的内涵；其次通过数据挖掘，剖析了人工智能伦理的多方利益相关者及其行为，界定了人工智能伦理边界及框架要素；最后结合理论和实践两个层面，构建了以人为中心、集设计伦理和社会伦理于一体的人工智能伦理框架。

Large language models represented by ChatGPT have promoted the formation of a new paradigm of artificial intelligence, while also arousing people's fears and concerns about artificial intelligence-centered information processing, and there is a growing outcry regarding ethical issues arising from artificial intelligence. Firstly, this paper systematically sorts out relevant literature on artificial intelligence ethics and conducts an integrated review of 115 officially released artificial intelligence guidelines worldwide, analyzing the connotation of artificial intelligence ethics. Secondly, through data mining, it analyzes the multiple stakeholders and their behaviors in artificial intelligence ethics, and defines the boundaries and framework elements of artificial intelligence ethics. Finally, combining theory and practice, it constructs a human-centered artificial intelligence ethical framework integrating design ethics and social ethics.