The purpose of this study is to analyze the educational needs of graduate stu-dents regarding artificial intelligence (AI) ethics and to examine whether these educational needs differ according to the frequency of AI use. Participants were 593 graduate stu-dents enrolled in a large university in Seoul, South Korea. We used a self-reported survey to measure students' levels of current understanding and their perceived importance of 22 items related to AI ethical standards. We analyzed the data using the Borich needs as-sessment model as well as The Locus for Focus Model. Our results identified safety, ac-countability, transparency, human rights protection, and non-infringement as the top pri-orities for AI ethics education for graduate students. Accountability emerged consistently and was identified as the top priority for educational needs, regardless of the frequency of AI use. On the other hand, privacy and transparency were found to be more important for those who use AI less frequently, and safety was found to be more important for those who use AI more frequently. Our findings will provide valuable insights for designing and developing AI ethics educational programs for graduate students, ultimately helping them in enhancing their ethical awareness and competence in relation to AI.