Since 2017, deepfakes have emerged across the internet. Deepfakes are audio clips, video clips, or still images that are intentionally constructed to deceive an audience into believing something that may not be true. This recent technological exploitation has brought with it a host of practical and ethical concerns. In 2020, corporations and scholars alike have come to realize that deepfakes are a critical issue that must be addressed. With so many facets of society wallowing in profound divisiveness, deepfakes are one more way to amplify and extend this unhealthy discord. This thesis examines the risk associated with deepfakes and whether this growing menace can be reduced or eliminated through a deepfake detection system. First, a common analytical framework for a deepfake detection system is presented. This framework creates a baseline set of components, including two new pieces yet to be considered. Second, a multisource case study methodology is used that evaluates multiple existing deepfake detection systems to examine how each fit within the architecture of the framework. Third, an analysis of the strengths and weaknesses of each system is performed to further provide validation of the research. The creation of a new deepfake detection system architecture can be used to develop new and improved systems to combat the growing danger posed by deepfakes. As the methods used to create deepfakes become more robust, so too must the technology tools used to combat these dangers.