Metal additive manufacturing is a rapidly developing high-efficiency material processing technology in recent years. In order to ensure the quality and reliability of fabricated parts, the formation of defects within them that have a significant impact on the mechanical properties of the structure should be avoided. In this paper, the application of laser infrared thermography for detecting subsurface defects in metal additive manufacturing is investigated. Firstly, based on finite element simulation results, the reliability of laser infrared thermography for detecting subsurface defects of different depths and sizes in metal additive manufacturing is studied. Furthermore, the influence of rough surfaces on detection is taken into account, and the noise suppression performance of commonly used infrared thermography sequence processing algorithms is comparatively verified. Finally, Experimental validation of the processing of selective laser melting specimens with artificial internal defects is carried out. The simulation and experimental results demonstrate that laser infrared thermography can reliably detect the internal sub-surface defects of metal additive manufacturing with a width-to-depth ratio greater than 1 and the spatial noise caused by rough surface interference can be effectively suppressed by the commonly used pre-processing method of thermography sequences. In addition, laser infrared thermography inspection is expected to be a reliable technology for online monitoring of metal additive manufacturing due to its advantages of high efficiency, non-contact and visualization.