Ethical thinking and reasoning is considered a core component of artificial intelligence (AI) literacy. However, there is a lack of strategies and assessments to promote and measure students’ ethical thinking in AI, particularly in K-12 education. In this paper, we discuss how RAICA, a project-based AI literacy curriculum integrates ethics into its framework and instructional resources. We employ a mixed-methods convergent design to obtain different yet complementary data on ethical thinking as an outcome mediated by diverse RAICA materials. Data analysis revealed that teachers utilized a variety of instructional strategies to foster students’ ethical thinking and that students actively engaged in ethical thinking activities, resulting in a developing understanding of stakeholders and potential benefits/harms of AI. Our work makes a key contribution to AI education by providing empirical evidence to support mechanisms for integration and assessment of ethical thinking within AI literacy curricula.