Deepfakes are manipulated or digitally generated images or video content using advanced artificial intelligence techniques. These generated contents are highly realistic and convincing, making distinguishing them from authentic sources difficult. Deepfake images have a profound effect due to their huge hazards in several areas, such as the spread of disinformation and false news, breaches of privacy, and cybersecurity threats. Detecting deepfake images is challenging because of the sophisticated nature of the technologies that continuously evolve, such as GANs, used to create high-quality, realistic, and closely mimicked genuine images, making them difficult to distinguish. To address this issue, we have proposed ImageShield, a sophisticated generalized deep-learning architecture using a convolutional neural network and integrating batch normalization techniques. To validate the efficiency and performance, we used two different datasets on the same architecture with the same parameter and were able to achieve 95% accuracy. We have also compared the ImageShield with some existing techniques and got slightly better performance with limited computational resources. The purpose of this study is to contribute to the continuing endeavors to combat the inappropriate use of AI in generating misleading visual information, thereby promoting enhanced confidence in digital media.