A fake image of President Barack Obama shaking hands with President Hassan Rouhani of Iran. A real photograph of a Muslim girl in an office doing her work with Donald J. Trump playing on a loop in the background of the television.
It's not always easy to tell the difference between real and fake photos. But the pressure has never been greater, as fake political news grows day by day.
On Tuesday, Jigsaw, a cutting-edge technology company owned by parent company Google, unveiled a free tool that researchers said could help journalists spot doctored photos — even those created with the help of artificial intelligence.
Jigsaw, known as Google Ideas when it was founded, said it is testing the tool, called Assembler, with more than a dozen news sites and organizations around the world. It includes Animal Politico in Mexico, Rappler in the Philippines and Agence France-Presse. It has no plans to offer the tool to the public.
“We’ve seen an evolution in how disinformation is used to manipulate elections, wage war, and disrupt society ,” writes Jared Cohen, CEO of Jigsaw, in a post for Assembler. “But as disinformation tactics evolve, so too do the technologies used to detect and ultimately stop disinformation.”
The tool is intended to verify the authenticity of images – or to show where they may have been altered. Journalists can upload images to the Assembler, which has seven «detectors», each of which is designed to detect a specific type of photo-manipulation technique.
When an image has been edited – for example, two images were merged or something was removed from the background – traces of the changes can remain behind. Assembler has been trained to be able to detect the relevant changes.
Five of Assembler's image detectors were developed by research teams at universities, including the University of California Berkeley, the University of Naples Federico II, and the University of Maryland. The models can detect things like color, areas of an image that have been copied and pasted multiple times, and whether more than one camera model was used to create an image.
“These detectors cannot completely solve the problem, but they represent a significant tool for combating misinformation”, said Luisa Verdoliva, professor at the University of Naples and visiting researcher at Google AI.
The other two detectors were developed by Jigsaw. One was designed to recognize the “deepfake”, realistic images that have been heavily manipulated by artificial intelligence in ways that deceive the public.
Santiago Andrigo, Jigsaw's product manager, said Assembler could be "very useful in a situation where a journalist from a major news organization receives a scandalous image and is under pressure to publish it immediately." It could also be used to verify an image that has gone viral, he said.
The database describes the “players” involved in influence, the common tactics used, and how lies were spread across social media platforms. Jigsaw worked with the Atlantic Council’s Digital Forensics Lab to curate the collection of nearly 60 disinformation cases, culled from over 700 investigations, articles, and reports published by the lab over the past five years.
Emerson Brooking, a resident at the lab, said the goal was not to build an encyclopedic catalog of disinformation campaigns, but to create a pattern by finding common elements in trying to describe the various disinformation efforts. In doing so, they could develop a list of elements that could help other media outlets and groups doing studies on disinformation, he said.
The two projects, the Assembler and the interactive misinformation platform, were announced in the new research edition of Jigsaw, The Current.
