This study examines acceptance of autonomous weapon systems (AWS) among military personnel and general citizens by distinguishing between support for their development and adoption versus their deployment and use. Using survey data from South Korea (general public, N = 1,222; military personnel, N = 155), we constructed hierarchical and interaction regression models to examine how demographics, value orientations, perceptions of just war and risk, and attitudes toward and knowledge of AI shape acceptance across different stages. Results show that military personnel report higher support for AWS development but substantially lower support for use than public. For development acceptance, the strongest drivers are interest in AWS and perception of just war compliance. For use acceptance, perception of just war compliance and acceptance of algorithm-based decision-making are the most influential predictors. Notably, knowledge of AI is negatively associated with acceptance of AWS use. Furthermore, this negative association is moderated by value orientations: individuals with stronger post-materialist priorities exhibit a steeper decline in support for AWS use with increasing AI knowledge, whereas materialist individuals demonstrate comparatively stable patterns. These results indicate that AWS acceptance is stage-dependent, conditioned by stakeholder responsibilities and value-based divisions. Consequently, governance discussions should differentiate between development and deployment/use stages, placing greater emphasis on operational risk and ethical legitimacy.