Deepfake technology uses advanced machine learning and AI techniques, particularly Generative Adversarial Networks (GANs), to generate hyper-realistic synthetic media. This advancement presents issues of privacy, security, and misinformation. Deep-fake technology is prone to significant threats in the digital age. Motivated by this background, detection of deepfakes with an appropriate methodology requires an in-depth systematic approach to learning. This work explores Convolutional Neural Networks due to their strong discriminative capability between the real and deepfake images. Additionally, feature extraction techniques have been used to efficiently differentiate real and fake images. Since deepfake images are generated using a sophisticated technology such as Generative AI, it becomes challenging to detect the fake images. This paper discusses the relevance of Convolutional neural networks for deep fake detection and the proposed work exhibits an accuracy of 91.67%. The proposed methodology firmly establishes that the application of convolutional neural networks is relevant and can be very effective for the identification of deepfake images.