Abstract

Abstract
AI transformation often fails not because systems stop functioning, but because organizations fail to sustain the routines, accountability structures, and interpretive capacities required to keep AI meaningful over time.AI systems are increasingly used to improve decision quality, operational coordination, risk management, and strategic renewal. Yet many AI initiatives remain technically operational while failing to become durable organizational capabilities. This paper argues that the central weakness in much AI transformation research is not the absence of implementation models, but the limited theorization of how AI becomes organizationally internalized and institutionally survivable after deployment. To address this gap, the paper develops a conceptual framework of AI survivability through enabling governance. It distinguishes AI adoption, AI implementation, and organizational internalization; reconceptualizes AI as a governance-embedded organizational capability; and theorizes enabling governance as the mechanism through which organizations reproduce alignment among AI systems, routines, accountability structures, managerial interpretation, and institutional learning over time. The paper introduces three linked concepts: digital implementation without organizational internalization, cumulative mediation, and organizational AI survivability. It argues that AI becomes sustainable only when organizations embed it operationally, structurally, and cognitively, and when governance repeatedly regenerates alignment under leadership change, restructuring, regulatory uncertainty, and technological disruption. The framework contributes to AI and organization studies by shifting attention from technical deployment to institutional durability, from restrictive governance to enabling governance, and from implementation success to recursive organizational reproduction. It also extends the societal relevance of AI governance by showing how weak internalization may undermine accountability, public trust, and the reliability of AI-supported decisions.

Keywords