Drag Your GAN is a sophisticated AI tool designed for flexible and precise control of visual content generation, specifically focusing on the manipulation of the pose, shape, expression, and layout of generated objects. Its main infrastructure relies on the use of generative adversarial networks (GANs), which are traditionally controlled via manually annotated training data or prior 3D models. However, this tool advances the field by introducing a powerful approach called DragGAN. DragGAN is unique because it allows users to ‘drag’ any points of an image to reach specific target points interactively, offering impressive flexibility, precision, and generality. Two main components form the essence of DragGAN. The first is a feature-based motion supervision that navigates the handle point towards the target position. The second component uses a novel point tracking technique that leverages the discriminative GAN features to continually localize the handle points position. Through DragGAN, users can deform an image with precise control over pixel movement, thus manipulating different categories, including animals, cars, humans, landscapes, etc. These manipulations are performed on the learned generative image manifold of a GAN, which tends to produce realistic outputs for even challenging scenarios such as hallucinating occluded content and deforming shapes. Both qualitative and quantitative comparisons show DragGAN’s superiority over traditional approaches in image manipulation and point tracking tasks. Furthermore, DragGAN also enables the manipulation of real images through GAN inversion.