Google Withdraws Satellite Imagery AI Tool Over Misinformation Risks

Google has halted a newly launched feature that enabled users to generate artificial images on top of real satellite and aerial photography within Google Earth. The company paused the tool less than 48 hours after announcement following reports of misuse and concerns from misinformation experts. The feature allowed text prompts to create false scenarios that inherited the credibility of genuine map coordinates and imagery.

What Happened

Google announced Nano Banana 2's integration into Google Earth on Thursday, positioning the capability as useful for educators, architects, and urban planners. The tool enabled users to zoom into locations and generate custom imagery by typing text descriptions—such as visualizing historical sites, planning real estate projects, or imagining neighborhood transformations.

However, within hours of the feature becoming available on the web version of Google Earth, users began creating and sharing screenshots of fabricated scenarios. A BBC investigation documented the creation of several false images: a collapsed Eiffel Tower, a sinkhole consuming the Great Pyramid of Giza, and Russian tanks positioned in Kyiv. The investigation also revealed users generated images of a non-existent Iranian nuclear power plant, a refugee camp on the US-Mexico border, and a fake Gaza hospital with bomb damage nearby.

The Core Problem

Misinformation researchers identified a critical vulnerability: the generated images were "welded to genuine coordinates, drawn on genuine imagery," according to AI expert Henk van Ess. This meant false content inherited the trustworthiness associated with Google Earth itself. As van Ess explained, the forgery did not need to be convincing in isolation because it drew credibility from the authentic map beneath it.

Google acknowledged this risk in its suspension statement, noting: "We know that people uniquely trust Google Earth for a reliable view of the world." The company observed violations of its content policies in user-generated screenshots before taking action.

Safety Measures and Limitations

Google stated that generated images contained invisible watermarks and could be verified using its Gemini chatbot or Lens search feature. However, BBC testing found these safeguards were not foolproof—researchers successfully circumvented detection systems and tricked Gemini into validating fake satellite imagery as authentic. External AI detection tools also failed to identify some Google Earth-generated content in testing.

The company said it is implementing "stronger guardrails" before reintroducing the feature. Google emphasized that generated images did not appear in the main Google Earth experience for other users without their direct involvement.