• Elevated pore pressure, modified fault-loading conditions, and aseismic slip are primary drivers of fault reactivation and induced seismicity.
• Probabilistic forecasting and machine learning are increasingly used for seismic risk assessment, but face limitations from geological heterogeneity and data scarcity.
• Fluid–rock interactions complicate the development of universally applicable models, constraining the generalizability of mitigation strategies.
• Advancements in understanding mechanisms and improving prediction are essential for safe and sustainable deep-subsurface energy development.