Navigating AI ethics in K–12 classrooms: exploring teachers’ AI ethics alignment through activity theory

AI ethics in K–12 education remains conceptually visible but practically ambiguous: while teachers are expected to use AI responsibly, concrete boundaries for ethical classroom use remain largely undefined. Drawing on Activity Theory, this study examines how K–12 teachers in China understand AI ethics alignment and how the mediating elements of the activity system shape these understandings. Based on metaphor elicitation tasks completed by 26 teachers and semi-structured interviews with 16 of them, the study reveals mixed and still-developing perspectives on AI ethics. Their AI ethics alignment was shaped through the interaction of rules, tools, community, and division of labor. Teachers translated vague ethical expectations into provisional classroom boundaries, made situated judgments about the risks created by AI tools in classroom practice, and often acted as frontline gatekeepers when shared community norms and responsibility structures remained underdeveloped. The findings call for explicit school-level AI ethics protocols, stronger institutional support for teachers’ ethical judgment, and a more coordinated responsibility structure across teachers, school management, and AI developers.