Accurate prediction of cutting forces plays a critical role in controlling surface integrity of products and extending tool life during precision machining processes. However, empirical and mechanistic models rely heavily on expensive equipment and extensive cutting experiments [1]. Although recent studies have incorporated finite element analysis (FEA) into AI models [2], the influence of material constitutive properties on simulation accuracy has not been sufficiently considered, and the generalization capability of AI models across different materials remains limited. To address these limitations, this study proposes an FEA-assisted cutting force modeling approach guided by material constitutive parameters and the friction coefficient at the tool–workpiece interface. Initially, a sensitivity analysis of the Johnson–Cook constitutive model was conducted to quantify the impact of individual material parameters on cutting forces under high strain-rate and high-temperature conditions. By identifying the dominant parameters governing force formation, the constitutive models for five specific steels—S25C, SKD61, SKD11, P5, and NAK80—were precisely calibrated. Building upon this foundation, the friction coefficient was refined across varying cutting conditions. These calibrated parameters and friction coefficients were then integrated into finite element simulations to generate cutting force data with high physical fidelity. Subsequently, a prediction model was established by synthesizing material properties, tool geometry, and cutting parameters. To further enhance the model’s generalizability, transfer learning was employed to extend its predictive capabilities to new materials, specifically S50C and SCM440. The results indicate that the proposed approach achieves a maximum prediction accuracy of 90% across diverse materials and machining conditions, demonstrating robust cross-material generalization. By utilizing material constitutive sensitivity as a core modeling strategy, this study significantly reduces experimental requirements while providing a physics-informed and practically deployable solution for cutting force prediction.