Recent case-study research on L2 student engagement with automated writing evaluation (AWE) feedback shows it is not simply a matter of understanding and implementing suggested changes but can instead be strongly influenced by individual and contextual factors. The present multiple-case study builds on this work by investigating the extent to which engagement supports learning from, and is influenced by trust in, AWE feedback. Six Mandarin L1 university students at different levels of English writing instruction participated, each providing a text they had written for a current course, which was submitted to the tool Grammarly for feedback. Screen-capture recordings, stimulated recalls, and interviews were used to collect information about how much of the feedback participants viewed, which of it they accepted, which they ignored or modified, and why they did so. Findings showed participants adopting a proofreading as opposed to a learning orientation toward the feedback, perhaps because most of it included specific corrections, thus requiring little work on the user’s part. Whereas previous studies attributed more successful engagement with AWE feedback to higher language proficiency, the present research showed different dimensions of trust in Grammarly’s feedback to be determining factors.