The field of vehicle safety is experiencing significant advancements in modeling and prediction capabilities. Recent developments have focused on improving the accuracy and efficiency of crashworthiness analysis, ball trajectory prediction, and vehicle collision dynamics. Graph-based surrogate models and neural networks are being leveraged to enhance the representation of complex 3D components and physical phenomena, leading to more accurate predictions and reduced computational demands. Decoupled dynamics frameworks and data-driven predictive control approaches are also being explored for vehicle rollover prevention and collision prediction. These innovations have the potential to substantially improve vehicle safety assessments and reduce the risk of accidents. Notable papers include: A new graph-based surrogate model, Recurrent Graph U-Net, which demonstrates great accuracy in predicting the dynamic behavior of vehicle panel components. A universal model combining differential equations and neural networks for ball trajectory prediction, which achieves high accuracy and strong generalization with minimal training data. A decoupled dynamics framework with neural fields for 3D spatio-temporal prediction of vehicle collisions, which accurately predicts vehicle collision dynamics with reduced computational effort. A model-free vehicle rollover prevention approach using data-driven predictive control, which maintains vehicle stability without requiring explicit system modeling.