Machine learning (ML) is instrumental in the detection of anomalous and event signals from the evolution of fault/fractured systems caused by natural and engineered processes in the subsurface. Critical to emerging ML methods is linking these signatures to the underlying physics of damage to determine if failure is imminent. This requires the integration of diverse datasets and robust computational methods to identify and capture the measurable signatures and physics of failure. Few computational approaches for simulating damage have been verified against readily available experimental data, or have been validated via blind predictions of data sets ahead of data acquisition. This session will highlight: (1) experimental and field data related to damage evolution (acoustic emission, microseismicity, displacements, fluid pressure, flow, geochemical signatures); (2) comparison of computational models and experimental/borehole/field data that highlight model strengths and weaknesses; and (3) announcement of a damage mechanics computational challenge on damage evolution in brittle-ductile materials.