The field of physics-informed machine learning is rapidly advancing, with a focus on developing innovative methods for modeling and predicting complex systems. Recent developments have centered around integrating physical laws and constraints into machine learning models, enabling more accurate and interpretable predictions. This has led to significant improvements in areas such as fatigue crack growth prediction, elastoplastic material modeling, and world model simulation. Notably, the use of physics-informed neural networks and conditional neural constitutive laws has shown great promise in capturing complex material behaviors and generalizing across diverse physical scenarios. Furthermore, advances in uncertainty quantification and mechanistic interpretability have enhanced the reliability and trustworthiness of these models. Some noteworthy papers in this area include: PC-NCLaws, which proposes a generalizable framework for elastoplastic material modeling, and PhysWorld, which presents a novel framework for learning physics-consistent dynamics models from limited real-world video data. These developments have the potential to transform various fields, from materials science and robotics to climate modeling and beyond.