Identification and classification of AI-created synthetic images

Synthetic data generation has improved to such an extent that humans are finding difficulty to identify photographs of human-made images. It really is not impossible that we need to consider reliability and authenticity in data. In this paper is a process with which to improve the identification of AI-generated images with computer vision methods. The latent diffusion process generated a synthetic dataset based on CIFAR-10 dataset to compare a set of synthetic images to a set of real photographs, while this process applied computer vision algorithms, but primarily focused on CNN2D for image classification. A modified version of CNN2D was operated to improve performance while achieving a remarkable accuracy of more than 95% (95.77%). This process produced a more trustworthy detection of synthetic images which supports a reality, that AI-generated media brings to light a true concern. Data integrity and trust is a potential emphasis for published digital content, with computer vision technologies being a strong focus.