"Steg.AI: Safeguarding Images with Clever AI Defense Against Manipulation"

Researchers from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed a technique called "PhotoGuard" to protect against AI image manipulation. PhotoGuard uses perturbations, invisible to the human eye but detectable by computer models, to disrupt an AI model's ability to manipulate an image. The technique includes two attack methods: an "encoder" attack that alters the image's latent representation, and a "diffusion" attack that optimizes perturbations to resemble a target image. By introducing these perturbations, the original image remains visually unaltered for humans but is protected against unauthorized edits by AI models. The researchers emphasize the need for collaboration among model developers, social media platforms, and policymakers to combat image manipulation.