In today’s world of digitization, the integrity and authenticity of the news is becoming very challenging. To address the issue, the authors in research work explored machine learning models for detecting fake face in real-time. Authors in this work presented four machine learning models comprising data collection, pre-processing, feature retrieval, model training, and evaluation. Support Vector Machine (SVM), K-Nearest Neighbours (KNN), Decision Tree (DT) Classifiers, and Convolutional Neural Networks (CNN) algorithms built using Siamese Networks are utilized to compare real face images with fake face images. These models are assessed based on accuracy, F1-score, and classification report metric. Objectives: The objective of the research work is to access the effectiveness of proposed models for the detection of real and fake faces in real time. Methods: Proposed models are assessed on the Kaggle dataset by detecting fake and real faces. For evaluation accuracy, the F1-score and classification report metric are calculated. Results: The SVM classifier achieved an accuracy of 58%, while KNN and Decision Tree classifiers obtained accuracies of 52% and 53% respectively. A Siamese Network model that retrieves features using MobileNetV2 attains a test accuracy of 63%. Furthermore, the outcomes show that the (CNN) algorithms built using the Siamese Network work better than alternative algorithms, achieve an accuracy of 94.5%, and exhibit significant proficiency in distinguishing between authentic and fake faces. Conclusion: After achieving experimental results it has been concluded that the CNN algorithm with Siamese Network performed well compared to other proposed algorithms. Considering all these things, the work PICSLEUTH helps to enhance the credibility of digital material and tackles the problems caused by the widespread use of deepfakes in this digital sphere.