Fake videos and photos are increasingly being used by cyber criminals for tricking people, and this problem is very concerned about the company Adobe, which is the developer of popular image editors. At the moment she, together with scientists from the University of California at Berkeley, is developing artificial intelligence to recognize fake photos, and the repeal of all amendments thereto — in the end developers want to get the original picture. A prototype tool to undo retouching has been established and is already showing interesting results.
The tool is based on the work of the neural network, which clearly sees the differences between edited and original photos. In order to teach it, the developers used the base with a pair of photographs, one of which has no patch, and the other is edited using the Photoshop Liquify for warping entities.
Tool Photoshop based on neural networks
This type of editing developers were especially interesting, because a couple of clicks of the mouse can completely change the facial expression of the person. Editing the photos so scammers can easily lead people astray, and this was proven in the course of the experiment. The researchers collected volunteers and asked them to guess the real picture in front of them or not. Special vigilance showed only 53% of people, and the rest did not find anything suspicious. Artificial intelligence Adobe, in turn, recognized fake with 99% accuracy.
The developers claim that artificial intelligence might even show how to edit fake to return to their original state. However, at the moment, his recommendations have produced mixed results.
The idea of a magical universal “cancel” button to return the image to its original state, is still far from reality. But we live in a world where it becomes increasingly difficult to trust we consume digital information, and I look forward to further exploring this area of research, — said Richard Zhang, the employee of Adobe.
To stop the spread of false news is extremely difficult
This is probably the first of its kind study in which scientists want to create a tool to detect fake photos. Unfortunately, even if such a tool is created, it will not be able to eradicate the spread of false news — even after being exposed, fake can still get into the social network and make people believe the wrong information.
Although we are proud of the impact that Photoshop and other Adobe products have had on the world, we also recognize the ethical implications of our technology. Fake content is a serious and increasingly urgent problem — shared Adobe in his blog.
Deception in social networks using a fake photo
One of the first cases of deception of Internet users with fake photos was described by the news Agency Associated Press. The social network LinkedIn was discovered the profile of a girl named Katie Jones to determine that under this account, there is potential spy managed a few. However, to recognize the fake photo was quite easy — it was created by means of neural networks and have had features, which you can read in our material.