Synthetic media, including deepfakes, represent a major threat in social engineering, as scammers increasingly use these technologies to impersonate individuals and commit fraud. The Synthetic Media Attack Framework is proposed to categorize these attacks based on five dimensions: medium, interactivity, control, familiarity, and intended target. While technology-based detection methods exist, this work emphasizes human-centric countermeasures, such as using proof-of-life statements and multi-factor authentication, to combat these threats effectively. The framework aims to help users recognize inconsistencies and implement security policies to protect against synthetic media attacks.