GAN stands for Generative Adversarial Network. It is a deep learning model where two neural networks (a Generator and a Discriminator) compete with each other. The Generator creates synthetic data, and the Discriminator distinguishes real and fake data, which leads to more  realistic outputs.