In today's world of Generative AI, it has become harder to tell the difference between real images and those created by artificial intelligence or carefully edited by professionals. However, many AI-generated images leave behind small patterns that act like a “mathematical fingerprint.” In this work, concepts from Linear Algebra and Fourier Series are used to detect these hidden patterns. Python is also used as a practical tool to uncover these glitches and help identify deepfake images. Images as Tensors: The process begins with understanding how images are stored in computers. Light and colors are represented as numbers and arranged into arrays using NumPy. This step shows why Linear Algebra plays an important role in computer vision. Matrix operations are applied to manipulate image data, and Singular Value Decomposition (SVD) is used to simplify complex images and extract their most important features. This helps reduce noise and highlight meaningful patterns in the image. The Frequency Domain: After working with images in their pixel form, the next step involves viewing images in a different way using the Fourier Transform. The Fourier transform is like a triangular prism that splits white light into the 7 known colours, however in this context it's waves or a compound wave being split into separate waves with different frequencies. Instead of focusing only on pixels, images are analysed as combinations of waves with different frequencies. This makes it easier to notice differences between natural images and AI-generated ones. Natural images usually show smooth and balanced frequency patterns, while generated images or deepfakes often reveal unusual frequency signals. Python Implementation: The final step focuses on applying these ideas in practice using Python. Libraries such as NumPy, SciPy, and OpenCV are used to perform Fast Fourier Transforms (FFT) on image data. The results are then visualised to reveal special patterns known as spectral spikes. These spikes act as indicators that an image may be artificially generated or manipulated, helping in the detection of deepfake images.