R&D > GANGRENE
R&D > GANGRENE
GANGRENE is a proof-of-concept steganographic capability based on an emerging way of training neural networks known as Generative Adversarial Networks (GANs). It is a medium-agnostic steganography framework for embedding messages in features rather than noise developed under internal research and development. For example, if the medium is a face, the GAN learns to embed messages by slightly altering the features of the face (e.g., opening the mouth, closing the eyes, changing the hair color).