AI can not only generate new photos, but also edit existing ones with high quality
(photo: CC0 Public Domain)
Image falsification manipulations have become more common since artificial intelligence came into mainstream use. Engineers from the Massachusetts Institute of Technology (MIT) propose two ways to protect images from deep forgery and editing using AI.
The AI-powered Dall-E and Stable Diffusion image generators are just the beginning of a new era in graphics processing. AI can not only generate new images, but also edit existing ones with high quality, leaving room for potential abuse in the form of deepfaking.
In response to this challenge, MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) has developed PhotoGuard technology, which makes it difficult for AI algorithms to alter images. The technology includes two methods to combat AI algorithms: “encoder” and “diffusion”.
The first method encoder adds a hidden representation of the protected image – the technology changes individual pixels in the image in a special way and prevents AI from recognizing the content of the image, meaning it blocks the ability of artificial intelligence to edit it.
The second, more advanced and resource-intensive method diffusion masks one image under another in the AI’s “eyes”. As a result, the artificial intelligence tries to change only the picture it “sees”, but does not touch the original, and in the end the image it generates looks unrealistic.