The Symbiotic Author as an Institutional Category Authorship, Creative Provenance, and Eligibility in Human–AI Literary Production

The Symbiotic Author as an Institutional Category extends a public research line begun with Symbiotic Dialectics: Meaning Emergence in Human–Non-Human Interpretive Systems (January 2026; DOI: 10.5281/zenodo.18187335) and developed through Symbiotic Dialogics (April 2026). That line moves from the analysis of sustained human–non-human meaning-making, to a literary method of human–AI dialogue, and here to a concrete institutional problem: how should institutions classify the human author when generative systems materially participate in the causal development of a work without displacing the human authorial locus? This preprint introduces the Symbiotic Author as a work-relative institutional category. The category does not depend on how much machine-generated language survives in the final text. Instead, it identifies a materially consequential creative relation in which system contributions receive substantive human authorial uptake and resulting human interventions, in turn, condition subsequent system contributions. The paper argues that four questions currently collapsed in debates over AI authorship must be kept separate: what creative relation occurred; what machine-generated material survives in the artifact; what evidence supports the account of the process; and what the institution permits. On that basis, it distinguishes textual provenance from creative causal provenance and separates authorship classification from disclosure, evidentiary corroboration, and institutional eligibility. The contribution is intentionally narrower than a claim to have invented human–AI co-authorship, distributed authorship, or the term “symbiotic authorship.” Those ideas have identifiable predecessors. The stronger claim advanced here is architectural: a work-relative human category, a causal-relational criterion, and a governance model in which classification, provenance, corroboration, and eligibility remain analytically independent. The paper develops formal criteria, boundary cases, an anti-anthropomorphism constraint, a materiality test, multi-author and ghost-operator cases, a proportionate corroboration model, and an operational protocol that can be adopted by journals, literary prizes, publishers, certification systems, archives, and other cultural institutions without requiring them to adopt the same normative policy toward generative AI. The relation may be symbiotic without the authorship being shared. Published as part of the research program of the Foundation for Relational Inquiry (FFRI-NM): relational inquiry without a predetermined answer.