Should AI-Generated Map Images Keep Their Warning After a Screenshot?

Google put Nano Banana 2 inside Google Earth on July 30. A day later it pulled the feature after people shared generated satellite scenes that violated policy. Google says the images were separate from the main map and carried SynthID. That protection got weaker on the route a suspicious image actually takes. Ars Technica photographed an altered image from a phone and could no longer verify it. BBC Verify found that outside detectors sometimes missed the images too. The clean export is the easiest copy to protect. The hard one is a crop dropped into a group chat during a fire, war or storm. If this feature returns, what should survive the screenshot—a permanent border, a visible label across the image, coordinates and prompt history, or all of them? And should images that can be mistaken for evidence be generated inside Google Earth at all?