Epistemic governance of AI augmented grounded theory: a validation gate framework for scaling qualitative inquiry

Purpose This paper develops a conceptual methodological framework for governing the use of artificial intelligence in the grounded theory. It addresses the risk that AI-generated outputs may be mistaken for qualitative evidence, interpretation or theory. Design/methodology/approach The paper synthesizes grounded theory traditions, computational grounded theory, large language model supported coding, human in the loop research, qualitative trustworthiness and critical AI scholarship. It develops a validation gate framework and uses a worked methodological vignette to show how AI suggestions can be documented, checked and bounded. Findings The paper argues that AI can support retrieval, organization, candidate coding, comparison and adversarial audit, but cannot authorize meaning, categories or theory. It identifies seven validation gates and four modes of AI use: assistive, dialogic, generative and adversarial audit. Originality/value The paper differentiates epistemic governance from reflexivity, trustworthiness, auditability and human in the loop design by focusing on the epistemic status of AI outputs before they enter the analytic record. It offers operational pass criteria for preserving theoretical sensitivity, reflexivity, auditability and researcher accountability when grounded theory is scaled through AI.