Google Pulls Google Earth AI Image Tool Over Fake Imagery

Google removed an AI image-generation feature from Google Earth on July 31, one day after its July 30 launch, saying some generated images appeared to violate its policies and risked undermining trust in the platform.

The feature was available on the web version of Google Earth and used an internal model called Nano Banana. After zooming to a location, users could click ‘create image’ and generate a satellite-style scene from a text prompt.

Researchers and journalists produced images that appeared to show events that did not happen. Examples included a blast crater in Los Angeles, protesters near a tech campus, a flooded U.S. Capitol and an island on fire.

Investigators reported that a range of prompts — including scenes of refugees at a border and a nuclear plant in another country — were accepted and produced imagery rather than being blocked.

In a post on X, Google wrote: “We know that people uniquely trust Google Earth for a reliable view of the world.” The company noted geospatial professionals had explored useful applications for the tool but that other users were sharing images that seemed to breach Google’s policies. Google disabled the feature on July 31 and stated it will add stronger protections before restoring it, without providing a timeline.

Google initially pointed to a SynthID watermark applied to every generated image to flag content as AI-created in its tools. Critics argued a visible watermark or embedded provenance tag would not prevent rapid sharing and viral spread of deceptive images because most viewers do not verify image metadata before reposting.

Google noted it already blocks creation of content on some harmful topics and said it would reinstate image generation only after implementing tighter guardrails. The company did not specify which policy violations had occurred or how many images were shared publicly, and it did not detail what additional safeguards would look like.

Satellite and aerial images have long been used by investigators, journalists and governments to corroborate reports because such imagery has been harder to fake. One-click generation of plausible, location-specific images introduces new verification challenges for those who rely on visual evidence.

News organizations, mapping professionals and governments are likely to monitor how Google strengthens protections and when the feature is restored.