Neural image compression employs deep neural networks and generative models to achieve impressive compression rates and reconstruction qualities compared to traditional signal-processing-based compression algorithms such as JPEG. However, color artifacts that arise in an image as the amount of compression increases have not been formally analyzed for neural-based compression algorithms. This paper provides an initial investigation into the degradation of color when images are compressed at comparable bit rates using lossy neural image compression and variants of JPEG. Our findings indicate that neural image compression degrades color more gracefully than JPEG, JPEG 2000, and JPEG XL.