Allostery is a fundamental process in regulating proteins’ activity. The discovery, design and development of allosteric drugs demand for better identification of allosteric sites. Several computational methods have been developed previously to predict allosteric sites using static pocket features and protein dynamics. Here, we present a computational model using automated machine learning for allosteric site prediction. Our model, PASSer2.0, advanced the previous results and performed well across multiple indicators with 89.2% of allosteric pockets appeared among the top 3 positions. The trained machine learning model has been integrated with the Protein Allosteric Sites Server (https://passer.smu.edu) to facilitate allosteric drug discovery.