There are times of certain insecurity on the Internet, a growing insecurity, especially due to the rise of artificial intelligence or neural networks, which are already able to create authentic images and faces so naturally that most people will find it very difficult to distinguish between reality and fiction. Now Google has managed to develop a tool with which it will be possible to detect images that have been artificially modified, so that we can know if it is reliable content and to be able to use it in publications that fight against Fake News.
An original way to detect modifications
It is not an application such as app that we can download from a play store. It’s a tool developed by one of Alphabet’s branches, the company holding to which Google has designed it to provide journalists and media. Nowadays it is very easy to deceive not only the media, but any person altering the appearance of an image. This is precisely what the “Assembler” tool pursues, which is what it is called, and now has its own website where it explains how it will work.
Jigsaw, the Alphabet company behind the project, has developed the tool to quickly detect any manipulation. This technology is helped by others that have been in the market for some time to fully analyze each image and give a verdict about its authenticity. For example, it is able to recognize pixel by pixel if they have been imported from other images or if they have been cloned by replacing the originals. It is also able to know if an image has been created with a tool developed specifically to create those known as Deepfakes, fake faces developed from AI over other existing faces.
From Jigsaw they recognize that the low resolution images are the great enemies of this tool, since in these images it is much more difficult to find those differences and alterations about the authenticity of an image. In fact, this tool usually works best when the image has its original size and retains its elements in the size with which it was conceived. To combat it, what this tool does is to use a reverse image search engine that always tries to find a similar image with more quality on the Internet. In this way, if you can reproduce the image with more quality and from another source you may be able to find the modification in a faster and more reliable way.
As we say it is a tool will be available for professionals who can know if an image is genuine or not before publishing it. Based on all the compared parameters of the image, the tool will return a score that will tell us whether or not the photo is original. In this way we seek to fight the Fake News, the Deepfakes and many other forms of visual fraud that we see more and more frequently on the Internet and on the networks.