Corporate tax compliance has significant implications for both public finances and corporate governance. Despite its importance, the prediction of corporate tax compliance remains a relatively underexplored area within corporate finance. Existing studies have predominantly focused on using artificial intelligence models, relying solely on financial indicators or combining these with other categories of information. This study introduces novel corporate governance drivers that leverage relational information through the board of directors’ networks. Using network analysis, six network features were identified to describe the company network, subsequently employed in developing AutoML models to predict corporate tax compliance. For this purpose, data from 4,969 companies were collected, and one-, two-, and three-year predictive models were created using H2O-AutoML. The AutoML approach was used due to the possibility of fully automating the ML development process. These multiclass models, incorporating two distinct network structures, were compared against models without network structures and described through accuracy, AUC-ROC or AUC-PR metrics. The empirical results did not reveal statistically significant differences between models utilising a board of directors’ network information and those without it. These findings necessitate further validation due to the limited number of tax compliance studies and the novel application of network analysis to board of directors’ relationships. Furthermore, the study aligns with Industry 5.0 principles by emphasizing integrating advanced technologies, such as AutoML and network analytics, to enhance decision-making processes in corporate governance and public finance systems.