With the all-round penetration of new AI technologies into human society, the necessity of policy regulation is becoming more and more prominent. A review of the research results on policy regulation of AI in the West from 1992 to 2020 reveals that: the research on AI policy regulation is still in its initial stage, and researchers mostly use law, political science and management science as their research perspectives to enter the topic, and the research fields are becoming more and more diversified, but not closely connected; The research topics are still at the level of philosophical value and contingency, specifically, the value judgment of policy regulation, procedural mechanism, and ethical issues of medical application, and there are controversies among different scholars, but a basic consensus has been reached on the ethical concerns and regulatory necessity of AI. The future research trends of policy regulation of AI mainly include three aspects: multi-discipline, implementation and diffusion of policy regulation, and ethical reflection.