Analogous to photosynthetic systems, photoactive semiconductor-based micro/nanoswimmers display biomimetic features that enable unique light harvesting and energy conversion functions and interactions with their surroundings. However, these artificial swimmers are usually non-selective and provide ineffective target recognition, resulting in poor surface analyte binding that affects the overall reactivity and motion efficiency. Here, the surface engineering of light-driven BiVO4 microswimmers by molecular imprinting polymerization is presented. After embedding surface recognition sites, the modified microswimmers can self-propel in a solution of a target molecule, without requiring toxic fuels, and degrade the target selectively in a pollutant mixture. These findings show that optimizing the design of semiconductor-based microswimmers with specific target recognition cavities on their surface is a promising strategy to achieve selective capture and degradation of organic pollutants, which is otherwise impossible because of the non-selective behavior of photogenerated reactive radicals. Moreover, this study provides a unique strategy to enhance the motion capabilities of single-component photocatalytic microswimmers in a specific chemical environment.