The company Jigsaw, owned by the Alphabet holding, announced the creation of a tool to detect fake photos. The new service will help identify traces of photo editing, even if they were created using artificial intelligence.
The project was called Assembler and was developed by scientists from the University of California, Berkeley and the University of Naples in Italy. According to the developers, it can make it easier to combat misinformation. The experimental platform consists of seven “detectors” operating on the basis of self-learning neural networks. Five detectors detect merging of photos or duplication of objects. The other two are designed to identify deepfakes.
“These detectors will not completely solve the problem, but they represent an important tool to combat disinformation,” said Luisa Verdoliva, a professor at the University of Naples.
According to Santiago Andrigo, product manager at Jigsaw, Assembler will be most important to journalists at major news outlets because it helps verify the authenticity of controversial images. It is reported that the free tool is already being tested in some public organizations and the media; the service will not be available to the general public.
Source: 3dnews.ru