Objective To explore the effectiveness of rolling out an advance care planning (ACP) training program guided by GROW model and supported by artificial intelligence technology among oncology nurses. Methods Using a whole-cluster randomized controlled trial design, 332 nurses were allocated to either a control group (n=165) or an intervention group (n=167). The control group received conventional training. The control group received routine ACP training. The intervention group received an ACP training program developed using the GROW model, comprising knowledge, tools, communication, and practice modules. Training was delivered over a 3-month period, supplemented by artificial intelligence-based support and multiple instructional formats. Pre-and post-training comparisons were made between the two groups on ACP knowledge, attitudes, self-efficacy, and attitudes toward death. Results After training, the intervention group scored significantly higher than the control group on ACP knowledge, attitudes, and self-efficacy. In the death attitudes scale, the intervention group scored significantly lower than the control group on death anxiety, death avoidance, and escape acceptance dimensions, and significantly higher on neutral acceptance and approach acceptance dimensions (all P<0.05). Conclusion The ACP training program based on the GROW model and supported by artificial intelligence significantly improves nurses′ ACP-related knowledge, attitudes, and self-efficacy and favorably modifies attitudes toward death.