The integration of Artificial Intelligence (AI) into organizational decision processes has produced a theoretically consequential paradox: as the availability of algorithmic recommendations increases, organizations often exhibit systematic patterns of operational inaction and responsibility avoidance. While modern organizations increasingly delegate operational choices to AI systems, the formal allocation of decision rights does not automatically guarantee effective decision enactment. Building on an enactment perspective of organizational architecture, this paper investigates how AI deployment induces authority ambiguity-a state where decision-makers experience uncertainty regarding their legitimate agency. In addition while existing research addresses this phenomenon through either a technological information-processing lens or a behavioral perspective on individual biases, both frameworks presuppose that informational sufficiency ensures decision enactment. We argue that this premise fails in culturally structured environments. To address this gap, we introduce a sociotechnical framework where the impact of AI disruptive properties (opacity, accountability displacement, and authority mimicry) on decision entitlement is fully mediated by the emergence of authority ambiguity. We argue that authority ambiguity acts as a focal proximal mechanism that hinders decision enactment. Furthermore, we position cultural infrastructure not as a direct causal determinant, but as a critical moderating boundary condition that shapes psychological safety and decision entitlement. By shifting the conceptual lens from deterministic mediation to a socio-technical moderating framework, we offer a more nuanced account of human-AI decision dynamics.