From Co-Author to Co-Algorithm: What Counts as “Authorship Evidence” in English Studies After Stylometry and Generative AI?

English Studies has long relied on a composite evidentiary logic for authorship attribution, material documents, stylistic judgment, institutional consensus, and computational stylometry without making that logic explicit. Generative AI does not invalidate computational stylometry but changes the conditions under which stylometric evidence can be interpreted. In historically secure corpora, textual surface may still function as an authorial trace; in contemporary contexts, AI-assisted drafting, synthetic style simulation, and contaminated corpora weaken the inference from surface to human source. The essay reconstructs a four-part warrant typology: material/documentary, stylistic/interpretive, institutional/contextual, and computational/probabilistic, tracing how generative AI exerts differential pressure across these categories. In response, it proposes the Hybrid Authorship Evidence Protocol (HAEP), a framework requiring provenance verification, corpus comparability, model transparency, adversarial testing, and disclosure of systemic limits. HAEP reframes authorship attribution as an argument genre with specified warrants, institutional thresholds, and acknowledged consequences, with implications for canon formation, pedagogy, and scholarly credit.