Who speaks when machines read scripture: boundary-making and interpretive authority in AI-mediated religious knowledge

Abstract This article examines how large language models increasingly mediate access to religious and culturally normative knowledge, where interpretive claims carry institutional and moral authority rather than merely informational value. It develops a boundary-making account of AI governance that treats interpretive authority not as a function of representational accuracy but as an outcome of sociotechnical practices across the AI pipeline, from the formalization of sacred corpora into datasets to model alignment and platform circulation. Drawing on science and technology studies, the analysis shows how these transformations shape both what meanings become legible and who is authorized to speak in the name of the text. The article introduces the Minimum Reporting Standards for Digital Analysis of Sacred Texts (MRS-DST) as a governance framework centered on disclosure across textual, algorithmic, and institutional domains, repositioning AI as an assistive analytic instrument rather than an interpretive arbiter in religious knowledge systems.