With the rapid growth of AI technology, notably in Generative Adversarial Networks (GANs), it is now possible to create incredibly realistic counterfeit images that are practically hard to differentiate from authentic ones with the naked human eye. This poses great challenges on assuring the authenticity of Social Media accounts and profile images. To cope with this issue, this research study presents an efficient and effective Deep Learning-based methodology to determine the presence of manipulations within the Twitter’s (X) profile image. The methodology leverages the advantages of combining ELA (Error Level Analysis) technique that helps to determine the discrepancy introduced during manipulation of image and Data Augmentation (DA) pre-processing techniques is to enlarge the dataset for improving the model’s ability to enhance the identification of legitimate and counterfeit images and Deep Learning-based ResNeXt CNN model to classify image as authenticate or fake.