["Human Factors and Ergonomics in Manufacturing &Service Industries, Volume 36, Issue 4, July 2026. ", "\nABSTRACT\nArtificial intelligence systems are increasingly deployed across diverse domains, and in some instances, their decisions may conflict with human ethical standards. Ethical dilemmas involving autonomous vehicle (AV) accidents, in particular, have attracted substantial scholarly attention. However, limited research has examined how users evaluate AVs when multiple morally salient features of a dilemma are considered together. To address this gap, this study investigates controlled trolley‐type AV dilemma scenarios by considering three factors: the decision‐making algorithm, the number of fatalities in a potential crash, and the user's relationship with individuals encountered on the road. We examine how these factors influence users' intention to use AVs through anger/disgust‐based negative moral affect, moral preference, and trust. Using structural equation modeling with data from 270 participants, we found that the number of fatalities significantly affected both negative moral affect and moral preference, whereas the relationship with individuals on the road influenced moral preference only. The decision‐making algorithm (DMA) did not significantly affect negative moral affect or moral preference, suggesting that users may rely more strongly on salient casualty outcomes than on abstract decision rules in AV moral dilemmas. Negative moral affect and moral preference shaped intention to use primarily through trust, with moral preference also exerting a direct effect. These findings contribute to AV ethics and acceptance research by identifying outcome salience and trust‐mediated moral evaluation as key mechanisms in users' responses to automated harm‐allocation decisions. Implications for ethically sensitive AV design and public communication are discussed.\n"]