Generative artificial 1ntelligence has quickly become a governance problem 1n higher education, yet much of the debate remains centered on cheating, plagiarism, detection, and academic 1ntegrity. This study shifts the focus from whether GenAI should be allowed to how course-level GenAI governance defines legitimate 1ntellectual labor and shapes graduate students’ emerging scholarly 1dentities. Drawing on a bounded 1nterpretive case study of one graduate-intensive college at a U.S. research university, the study analyzes twenty-two graduate course syllabi and eleven semi-structured 1nterviews with doctoral students. The analysis 1dentifies four course-level governance regimes: prohibitionist 1ntegrity, conditional authorization, assistive-generative boundary protection, and 1ntegration-as-pedagogy. Across these regimes, syllabi functioned as governance artifacts that translated 1nstitutional uncertainty 1nto local expectations about authorship, disclosure, responsibility, and acceptable assistance. 1nterview findings show that graduate students did not merely comply with or resist these expectations. They 1nterpreted AI use through questions of authorship, scholarly 1ndependence, linguistic legitimacy, reputational risk, and 1ntellectual credibility. Multilingual students, 1n particular, described GenAI as both a form of linguistic support and a possible source of stigma. 1ntegrating policy enactment, governmentality, Bourdieu’s symbolic power and linguistic capital, and boundary-work theory, the study argues that GenAI governance 1n graduate education moves from academic 1ntegrity to scholarly subject formation. The findings suggest that more coherent graduate AI governance should provide clearer course-level guidance, discipline-sensitive expectations, transparent disclosure norms, AI literacy 1nstruction, and protections for multilingual writers.
Evelyn Wu (Thu,) studied this question.