Extending the lifetime of aqueous batteries is critical to enable practical applications. It is currently limited by chemical reactions occurring in the electrolyte in uncontrolled ways. Using computer simulations, our aim is to understand how and why these reactions happen. While standard techniques limit us to only a handful of reactions, we propose to deploy a machine learning approach to accelerate our simulations. This project will allow us to achieve a microscopic understanding of aqueous battery ageing – in addition, we could provide guidelines to experimentalists to design electrolytes with improved performances.