Journal of Multimedia Processing and Technologies Sonal Lakade, Shyam Khairkar, Praveen Kokane, Amol Kale, Rajivkumar Mente 16 3 2025 https://doi.org/10.6025/jmpt/2025/16/3/113-137 https://www.dline.info/jmpt/fulltext/v16n3/jmptv16n3_1.pdf The rapid advancement of developments in artificial intelligence (AI), particularly in Generative Adversarial Networks (GANs), has paved the way for the creation of extremely realistic synthetic images that pose a challenge for digital forensics. Traditional image authentication techniques lack the pace to catch up with the growing sophistication of AI-synthesised images, calling for more innovative detection methods. This review examines the potential application of deep learning technologies, specifically Convolutional Neural Networks (CNNs), in detecting AI-generated synthetic images. The paper discusses several conventional and deep learning based approaches, compares their performance, and indicates the significance of lightweight CNN models in maintaining computational efficiency without sacrificing accuracy. In addition, it elaborates on the implications of explainable AI in bringing transparency to detection models. The review also explores the importance of synthetic image data in computer vision tasks, as well as its challenges, including domain gaps and biases. Lastly, the application of digital forensics in preventing misinformation and malicious acts involving synthetic images is discussed. The results underscore the importance of reliable and interpretable deep learning based detection methods in preserving the integrity of digital forensic examinations.