Workaholism is associated with dependency on Large Language Models in a cross-national study

The rapid adoption of Large Language Models (LLMs) has created new forms of digital reliance, yet little is known about how work-related pressures may be associated with this dependency. This study investigated whether two core dimensions of workaholism, working excessively and working compulsively, are linked to instrumental and relational dependency on LLMs across three national samples. Participants from China (n = 563), Germany (n = 360), and the United Kingdom (UK) (n = 567) completed validated measures of workaholism and LLM dependency. Configural and metric invariance were supported for both scales, enabling comparisons of associations across countries. Working compulsively showed consistent positive associations with both forms of dependency in China and Germany, with a weaker pattern in the UK. Working excessively was largely unrelated to dependency in simple correlations, although pooled regression models indicated small negative associations in the German reference group. Cultural moderation emerged for only one pathway: the link between compulsive work and relational dependency was significantly weaker in the UK than in China and Germany. Pooled models confirmed that working compulsively was the most reliable predictor of both instrumental and relational dependency, whereas working excessively showed modest negative associations. Chinese participants reported higher levels of instrumental and relational dependency than Germans; Chinese and British participants also showed higher instrumental dependency. These findings suggest that compulsive work habits make employees particularly susceptible to both instrumental and relational dependency on LLMs. For individuals exhibiting these patterns unrestricted access to LLMs may reinforce unhealthy levels of work involvement, hence increasing the likelihood of blurred work-life boundaries.