The rapid growth of technology has led to an alarming rise in the production and dissemination of fake videos, particularly in the context of war footage. This study addresses the critical challenge of detecting war footage forgery attacks applied to a custom dataset comprising both authentic and manipulated war-related content. Two distinct feature extraction methods, EfficientNet and Local Binary Pattern (LBP), were rigorously evaluated to assess their effectiveness in determining the authenticity of war footage. Machine learning models were employed to achieve an accuracy of 81.00%, demonstrating the viability of the proposed framework. The study contributes to the extensive conversation on combating the spread of propaganda and misinformation during times of conflict, ultimately fostering a more accurate understanding of events on the ground.