<p>Corporate greenwashing, where companies exaggerate their environmental actions, is a growing issue. Yet existing research on greenwashing primarily focuses on causal relationships, neglecting the comparative significance of various factors, and lacks flexibility and comprehensiveness in predicting. This study addresses these gaps by employing machine learning algorithms to analyze high-dimensional micro-level data from 5,245 firm-year observations of heavily polluting Chinese Listed companies between 2010 and 2021. We identify key drivers of corporate greenwashing through the best-performing model, using Shapley additive explanations to interpret results across four established causal dimensions. We find that LightGBM algorithm, based on a gradient boosting framework and tree-based learning, exhibits the highest predictive accuracy. National environmental regulation, a stringent control over companies, stands out as the most influential. Moreover, greenwashing decisions are primarily influenced by the board of directors rather than top management team, and factors like digital transformation and media oversight channels are proved less significant than anticipated. Some complex, nonlinear effects of various factors on greenwashing are also identified in univariate analysis. The core innovation of this study lies in identifying key factors and critically evaluating those established by existing research, thereby challenging their significance, while also developing a more precise and comprehensive predictive model. Based on the findings, we recommend policymakers enforce specific, targeted, and mandatory regulations, ensuring downward implementation to curb greenwashing.</p>