<p>Carbon fiber-reinforced polymer (CFRP) composites exhibit complex thermo-mechanical coupling effects during machining due to their heterogeneous and anisotropic nature, which easily induces microscopic failure behaviors and severely affects the integrity of machined surfaces. To accurately characterize the damage evolution of CFRP during machining, this study proposes a thermo-mechanical coupled orthogonal cutting modeling method for CFRP based on the modified Tsai-Wu criterion and Johnson–Cook criterion. A novel modified criterion considering the thermal softening effect is proposed, and the mechanisms of material stiffness degradation and element deletion are implemented through the VUMAT user subroutine in Abaqus/Dynamic, Temp-disp, Explicit modules in ABAQUS software. Combined with the proposed modeling theory and key process parameters, a cross-scale finite element (FE) simulation model capable of reflecting thermo-mechanical coupled damage effects under multiple working conditions is constructed to describe machining damage evolution behavior of CFRP. A series of orthogonal cutting experiments are performed on a self-constructed comprehensive experimental platform under identical conditions. Measured data of cutting force and temperature were used to establish mapping relationships with preset cutting speed and depth. The results demonstrate that the established cross-scale model can effectively capture the dominant failure modes under different fiber orientations. The maximum deviation between the predicted and experimental values of cutting force and temperature is less than 9%. The established polynomial fitted models can accurately predict the variation of cutting force and temperature with machining parameters, which provides theoretical support and a basis for process optimization of high-quality and efficient machining of CFRP components.</p>