When AI becomes indispensable: AI dependence as a pathway from perceived organisational AI adoption to employee withdrawal behaviour

Organisations are rapidly integrating artificial intelligence (AI) into work processes as a productivity-enhancing transformation, yet emerging evidence suggests the same technologies may elicit unintended employee responses. Although prior research has documented AI’s attitudinal consequences – such as job insecurity and anxiety – the behavioural pathway through which adoption signals become observable disengagement remains underspecified. This gap is theoretically consequential because withdrawal behaviour represents a quieter form of disengagement that may erode organisational functioning before it surfaces in standard metrics. Drawing on social cognitive theory, we examine whether perceived organisational AI adoption is associated with employee withdrawal behaviour, whether AI dependence – conceptualised as employees’ experienced psychological reliance on AI for task accomplishment – mediates this relationship, and whether proactive personality moderates it. Using three-wave time-lagged data from 328 full-time Korean employees and latent structural equation modelling, we find that perceived organisational AI adoption was positively related to withdrawal behaviour, that AI dependence partially mediated this relationship, and that proactive personality attenuated both the AI adoption–dependence link and the indirect pathway to withdrawal. These findings advance the conditional turn in AI-at-work research by identifying AI dependence as the integrative cognitive mechanism linking adoption to disengagement, and proactive personality as the dispositional buffer.