White spots in CycleGAN outputs can happen due to early training stages, hyperparameter issues, or lack of diverse training data. Keep training to see if it resolves, and check learning rates or regularize the discriminator. Using a black screen background can help spot display issues like white spots or artifacts more effectively, making it easier to address them.