Deepfake Satelite Imagery warns Geographers

When we think of Deepfake, we usually think of artificial intelligence-generated humans. This may be amusing, such as Tom Cruise, or nefarious. Deepfake geography, or AI-generated photographs of cityscapes and countryside, is something we don’t think about. However, this is precisely what some researchers are concerned about.

Geographers are particularly concerned about the spread of phony, AI-generated satellite imagery. Such images have the potential to deceive in several ways. They may be used to create wildfire or flood hoaxes, as well as to debunk reports based on actual satellite imagery. Deepfake geography may also be a national security problem, as geopolitical foes are misled by fake satellite imagery.

In 2019, the US military issued an alert about this very possibility. Todd Myers, a National Geospatial-Intelligence Agency researcher, imagined a scenario in which false data indicates a bridge in the wrong place, fooling military planning software. Myers said, “So from a tactical perspective or mission planning, you train your forces to go a certain route, toward a bridge, but it’s not there. Then there’s a big surprise waiting for you.”

According to Bo Zhao, an assistant professor of geography at the University of Washington, the first step in addressing these issues is to make people aware that there is a problem. Zhao and his colleagues recently published a paper on the topic of “deep fake geography,” which includes their experiments with creating and detecting such imagery. Rather than treating Deepfake as a novel task, Zhao and his colleagues situate the technology within a millennia-long background of fictitious geography.