Owing to layer-by-layer processing, additive manufacturing (AM) provides high flexibility in manufacturing components allowing virtually infinite geometric freedom. The use of AM and its particular realization – laser powder bed fusion – is rising. However, the method is not free from flaws, mainly represented by structural defects of the printed specimen, such as cracks and pores, requiring processing monitoring. Using a structure-borne high-frequency sensor makes it possible to detect acoustic emission (AE) associated with cracking, which occurs not only during the printing process but also after its completion. We discuss the machine learning approaches elaborated by our group aimed at in situ crack detection by the AE analysis.