The development of sophisticated algorithms that reliably distinguish between real and synthetic human faces has become necessary because of the rising concern over security breaches, particularly in sensitive applications like surveillance and access control. This project offers a cutting-edge security solution that uses convolutional neural networks (CNNs) to accurately identify authentic and phony faces. After improving the dataset with data augmentation techniques, a baseline model was created using a sequential CNN structure. To maximize efficiency, the model received extensive preparation, which included compilation and topological structuring. Following training the CNN model’s results showed a 93.91% test accuracy, a 0.1883 test loss, a 97.33% train accuracy, and a 0.0729 train loss. A Streamlit interface for users was created to enable real-time face identification allowing the model to be used for real-world security system applications in a smooth and effective manner.