The corrosion sensitivity of polycrystalline materials largely depends on their structural characteristics. Intergranular corrosion, as a common type of corrosion, exhibits stronger penetration and directional path characteristics compared to grain corrosion; the grain structure significantly influences the evolution process. Rapidly developing intergranular corrosion can weaken grain connections, ultimately leading to grain detachment, which causes a sudden increase in local corrosion depth. This, in turn, causes local stress concentration on the steel in engineering structures, reduces the number of cycles required to initiate local fatigue cracks, and poses significant hazards. The objective of this study is to investigate the grain detachment and corresponding corrosion evolution laws caused by intergranular corrosion under the influence of material structure. This is fundamental for accurately simulating corrosion and assessing the corrosion resistance of steel, which is currently lacking in existing research. This study is based on two types of grain structures: those randomly generated by the Voronoi method and those identified from metallographic experimental images. Using the cellular automaton method to investigate grain detachment behaviour under intergranular corrosion and analyse the overall corrosion evolution laws, including the temporal evolution characteristics of the statistical patterns of the corrosion front. This study incorporates grain detachment due to corrosion into the numerical simulation process to analyse the evolution of metal corrosion laws. It also examines the accuracy of representing actual intergranular corrosion patterns using grain structures generated by the common Voronoi method. This research extends the existing studies on intergranular corrosion and provides modelling suggestions for numerical simulations of intergranular corrosion.