In response to a high incidence of false information during the U.S. election season, Alphabet, Google’s parent company, has launched an AI tool called Assembler to help people verify that the original photos have been tampered with. It combines a series of tests to troubleshoot clues, such as seeing if a photo has been trimmed and pasted into another image, editing brightness, or removing certain areas from the background.
(From: FB, via Cnet)
For news outlets that need to be careful, Assembler may be a powerful fact-checking tool, such as Mark Zuckerberg, Facebook’s chief executive, who has long been suspected of having a robot-like fake laugh.
The Assembler project team is made up of many of Alphabet’s people, including engineers, designers, researchers, and policy experts, working to address issues such as false information, harassment, censorship, violent extremism, and election rigging.
With the development of deep learning and artificial intelligence technology, photo/video forgery tools such as DeepFake have attracted great attention. In today’s conflicting online environment, false information can easily spread through social networks.
The industry has made many efforts to address concerns, including explaining the details of technology, scrutiny, protecting the right to freedom, and reducing fraudulent content.
In a blog post Tuesday, Project Leader Jared Cohen said:
The Assembler tool has been used by news and fact-checking agencies including AFP, Animail Politico, Code for Africa, Les D?codeurs du Monde and Rappler.
For now, however, it is not possible for ordinary people to fact-check suspicious photos uploaded to Facebook.