Computer Vision techniques are extensively utilised in entertainment, enhancing realism in games and movies. Within video games, these techniques enable the recognition of objects, characters, and player movements, thereby enabling more intelligent game responses to player actions, resulting in more prosperous and more immersive experiences. Moreover, integrating deep learning methodologies with computer vision facilitates the automatic generalisation of special effects and enhances live broadcasts with interactive elements. However, such approaches also present challenges, as they can be exploited to manipulate and fabricate information, such as swapping faces in images or videos. This phenomenon is particularly evident in social media, where various forms of counterfeit or manipulated content proliferate, commonly called fake news. Our article proposes a novel convolutional neural network-based method for detecting alterations in real facial images and distinguishing them from artificially generated ones. Our technique, with its potential to revolutionise the field, enables the classification of facial photos into three categories: authentic faces, authentic faces with applied modifications (e.g., through photo editing software), and artificially generated facial images.