Ensuring long-term integrity of copper canisters is central to the safety of high-level nuclear waste disposal systems. In the Swedish repository concept, copper serves as the primary corrosion barrier; its failure would have severe environmental and societal consequences. However, under deep geological conditions, copper is exposed to groundwater containing multiple aggressive species (H, S, O, and Cl), whose coupled interactions can fundamentally alter corrosion pathways. Despite decades of research, the combined effects of these species, particularly their role in internal corrosion and grain boundary degradation, remain unresolved, representing a critical gap in the scientific basis for safety assessment. This project aims to develop a computational framework combining density functional theory (DFT) and machine learning to investigate multi-species corrosion mechanisms in Cu-based systems. Approximately 130,000 DFT configurations will be generated for Cu–O–S–H–Cl systems to train a high-accuracy interatomic potential, enabling large-scale molecular dynamics simulations of diffusion, segregation, and structural degradation. The generated data will include atomic structures, energies, and forces, and will be used for model development and analysis. This study will provide fundamental insights into copper corrosion under repository-relevant conditions and support predictive modelling for nuclear waste containment.