MIT's new tool 'PhotoGuard' to help against AI manipulation

While the techniques are effective, they aren't foolproof yet. (Image: MIT CSAIL)

The Computer Science and Artificial Intelligence Laboratory (CSAIL), part of Massachusetts Institute of Technology (MIT) in the United States has come up with a way to protect your photos from being altered with AI.

Called PhotoGuard, the tool prevents misuse of your photos by altering pixels in an image in a way that makes it difficult for the AI to understand. Called "perturbations" they are invisible to the human eye but can still be read by AI, reported Engadget. 

This comes amid rise of generative artificial intelligence (AI), which runs the risk of our photos being manipulated and used against us. We have already seen this in action, from Pope Francis wearing a puffer jacket to a fake Mark Zuckerberg spouting spiel on how he controls the world.

One way to implement this is by using an "encoder" method which introduces complications in information that describes the position and colour of pixels in an image. This alteration prevents the AI from understanding what it is looking at.

The second method known as "diffusion" attack disguises the image as another one in the eyes of an AI. This is done by taking information from another target image and altering the pixels in the source to resemble that, which masks the image in the eyes of the AI.

Speaking with Engadget, Heidi Salman, lead author of the paper and a doctorate student at MIT, said that a, "collaborative approach involving model developers, social media platforms, and policymakers presents a robust defense against unauthorized image manipulation".

While the techniques are effective, they aren't fool proof yet. A malicious actor could still reverse engineer the image by cropping, flipping or adding digital noise to it.

Salman admitted that, "much work is needed to make this protection practical. Companies that develop these models need to invest in engineering robust immunizations against the possible threats posed by these AI tools".