The potential of deep fake videos to be employed for malicious intents, such as disseminating false information, defaming individuals, and engaging in political manipulation, should not be overlooked. Nonetheless, it is worth acknowledging that deep fake technology can also have positive applications across various domains. Nonetheless, it remains crucial to acknowledge the potential risks associated with deep fakes and implement measures to curb their misuse. These measures encompass the development of effective detection tools for identifying deep fake videos, public education on recognizing and mitigating their impact, and the establishment of regulations to prevent their malevolent utilization. In this study, we successfully identified deep fake using pre-trained models based on CNNs. DenseNet121 and other well-known pre-trained models had their hyperparameters adjusted by us. The 70 k fake faces chosen from the one million fake faces (made by StyleGAN) and the 70 k real faces from the Flickr dataset collected by Nvidia supplied make up the deepfake dataset.