The study focuses on how Twitter facilitated the propagation of deceptive images during Hurricane Sandy. It reveals that a small number of users were responsible for most retweets of fake image tweets and that automated classification models were remarkably effective in discerning between authentic and fraudulent images based on the content of the tweets. The influence of follower relationships on the dissemination of these images was negligible. With the aid of classification models, the research achieved a remarkable accuracy of 97% in differentiating between fake and genuine images, which underscores the superiority of tweet-based features over user-based features in detecting false content.