Google rolls back AI image generator in Google Earth after misuse concerns

Google Earth’s web version introduced a feature this week that allowed users to generate images from text prompts using the platform’s satellite, aerial, and 3D imagery. The tool leveraged Google’s Nano Banana image generation model and was positioned by Google as useful for visualizing historical sites or real estate projects.

Within days, concerns about misuse emerged after examples surfaced showing AI-generated images of contested geopolitical scenes, including what appeared to be refugees near the Mexican border and a bomb crater near a hospital in Gaza. Google responded by noting that every AI-generated image in Google Earth includes the SynthID digital watermark, which users can verify via the Gemini app or Google Lens in Search.

However, experts demonstrated that the watermark is not foolproof. Henk van Ess of Digital Digging showed that AI-altered videos derived from Google Earth imagery could bypass Hive’s AI detector, highlighting practical limitations of detection mechanisms. Van Ess also recommended cross-checking suspicious images against other satellite platforms such as Sentinel-2 or Landsat, and verifying orbital metadata like capture dates and satellite sources.

Google confirmed in an update to the original article that it was rolling back the feature. The company did not provide additional details about the rollback timeline or whether a revised version would be reintroduced in the future.