Google has been forced to abruptly withdraw a new AI feature from Google Earth that allowed for the creation of fabricated satellite imagery superimposed on real-world coordinates. The move follows a backlash from Open Source Intelligence (OSINT) researchers who discovered the tool could be used to generate highly misleading content.
Threatening Geospatial Credibility
The tool, powered by Google’s Nano Banana 2 model, enabled users to add elements such as bombed neighborhoods, refugee camps, or nuclear facilities onto authentic maps. Researcher Hank Van Ess noted that the primary danger is that the forgery "inherits the credibility of the map on which it was created."
During testing, the AI fulfilled requests to depict the White House in a war zone with "blood and dead soldiers," while other users generated images of missile bases in Cuba and bomb craters in the center of the Kremlin.
The Failure of Safety Mechanisms
Despite Google’s claims that generated images include the SynthID digital watermark for identification, practical tests suggested otherwise. Tal Hagin, head of the OSINT platform Golden Owl, found that even Google’s own chatbot, Gemini, could not definitively confirm whether an image was AI-generated.
Google announced the temporary withdrawal of the feature, stating it is working on implementing stronger security mechanisms. The company acknowledged that the public places a unique level of trust in Google Earth as a reliable digital record of the world.
