As the seismic data obtained during hydrocarbon exploration today are significantly increasing, manual interpretation is getting longer, and study is being increased. For instance, up to one terabyte of data can be produced daily by a single seismic survey, and seismic data sets may exceed many petabytes quickly. In the last decade, interpreters have been using computer applications to accelerate the interpretation process. Research employing machine learning (ML) is being actively conducted in the petroleum industry in recent years. This study reviewed research papers published over the past decade that discuss ML techniques for fault detection and interpretation. The research trends and machine learning models explored in the 79 articles were studied in depth.