Google shuts down AI image-editing feature in Google Earth after one day

Google Earth’s experimental AI image-editing feature, which allowed users to alter satellite images using text prompts, was disabled within 24 hours of its launch. The tool used Google’s Nano Banana 2 model to generate AI-altered versions of real-world locations based on user input.

The feature was demonstrated by Digital Digging’s Henk van Ess, who generated images showing additions like refugees near the Mexican border and a bomb crater by a hospital in Gaza. Google initially responded that generated images included digital watermarks and that harmful topics were prevented, but later acknowledged that users shared screenshots of generated imagery that appeared to violate policies.

Google stated it was rolling back the feature to implement stronger guardrails, emphasizing that generated images did not appear in the main Google Earth experience for others and were watermarked as AI-generated. The company also noted that geospatial professionals had used the feature for useful purposes.

Van Ess reported that his testing showed the generated media could bypass AI detection tools, including fooling Hive’s AI detection system with a video generated in Google Earth. He stated that no prompts were refused, softened, or suggested for modification during his testing.