Google Earth adds genAI to fabricate satellite imagery

Google launched a new generative AI layer in Google Earth, powered by the Nano Banana model, enabling users to type text prompts and insert fabricated visuals—such as nuclear plants or bomb blasts—directly into the satellite map interface. This undermines the utility of Google Earth as a source of truth for open source intelligence (OSINT) analysts and journalists, as the generated content creates photorealistic depictions of events that never occurred. While Google asserts that SynthID watermarks allow for later verification and that filters prevent the most harmful prompts, security researchers emphasize that social media dynamics allow fake images to cause significant damage before verification occurs. The release effectively arms disinformation campaigns with a free, easy-to-use tool to create convincing geospatial forgeries.