Abstract: Metal-organic frameworks (MOFs) have garnered significant attention in the field of photocatalysis in recent years due to their inherent porous structures, tunable metal sites, and ligands, which confer properties such as high specific surface area, adjustable dimensions, and abundant active sites. However, the wide bandgap, low photogenerated carrier lifetime, and poor photochemical stability of MOFs limit their application in photocatalysis. Existing research indicates that modifying MOFs can significantly enhance their photocatalytic properties. This review discusses two key modification approaches for MOFs: intrinsic modification and composite modification. Intrinsic modification encompasses lattice distortion, macroscopic morphological changes, and elemental doping methods. Composite modification primarily covers techniques such as constructing heterojunctions, forming Schottky junctions, and introducing light-absorbing centers and electron-withdrawing centers. Modifying MOFs can enhance catalytic activity by optimizing the light absorption range, prolonging the lifetime of photo-generated carriers, promoting electron migration, and increasing the number of active sites, while preserving the porous nature of MOFs. This provides an ideal platform for the diffusion, adsorption, and reaction of the target substances. Additionally, this review reports on recent work utilizing machine learning methods to enhance the efficiency of MOF screening and predict their photocatalytic performance, representing a future trend in the development of MOFs as photocatalytic materials.