Abstract: The new round of technological revolution, represented by the research, development and application of artificial intelligence (AI) technology, has brought complex ethical issues and governance challenges while improving people's lives. In order to address these issues and challenges, this paper analyzes the nature of the AI ethical risks from the dimensions of algorithm, data and application. The paper also surveys typical ethical risks from various case studies, and proposes a sustainable risk management approach. Based on the approach, the paper makes several policy recommendations on how to manage the ethical risk in AI.