Institutional warrant and the editorial signature: AI on both sides of scholarly production and discovery

Purpose I hypothesize institutional warrant as the consumption-side analog of the editorial signature in AI-reorganized scholarly communication, and I situate the hypothesis within a century of warrant theorizing in library and information science (LIS). Design/methodology/approach The paper offers a conceptual analysis linking. Floridi’s AI-born journal proposal, Chavez Moutran’s Model Context Protocol discovery prototype, Sundin’s account of source invisibility, Bates’s division of labor in search interfaces, the LIS warrant tradition from Hulme to Martínez-Ávila and Budd, Gazit’s theory of constitutive knowledge sources and Van der Veer Martens’s grounding of ontic trust. Findings AI compresses, through its machine efficiencies, procedures on both the production and the consumption sides of scholarly communication, but legitimacy depends on preserving locatable human and institutional judgment. Four design commitments directly follow: load-bearing provenance, selective friction, archival-mode article search and visibility of production-side conditions. Originality/value The paper reframes library discovery as a constitutive institutional practice rather than a neutral access layer and argues that warranted openness strengthens fair use, open access, open science and open literature.