Google pulls AI image generator from Earth app after misinformation concerns

Google said on Friday it will remove a newly introduced AI image-generation feature from its Google Earth platform after users produced realistic but fabricated satellite images that raised misinformation worries. The tool, built on Google’s Nano Banana 2 technology, allowed users to type a text prompt and generate photorealistic images for any location in the service’s satellite, aerial and 3D views.

Within hours of its launch on Thursday, users posted examples that added fictional bomb blasts, refugee camps, nuclear plants and other scenes to real-world locations. An Open-source intelligence researcher who tested the feature said the system did not block any of the prompts he tried, including “refugees near the Mexican border” and a “nuclear plant in Iran.”

German investigative reporter Ben Huebel described the feature as “absolutely irresponsible” in a LinkedIn post, warning that it could produce “damning images” resembling conflict zones and complicate verification efforts. Google’s own press release had promoted the tool as a way to “visualize history, create real estate plans and more,” suggesting it could help users imagine changes to their neighborhoods.

In response to the backlash, Google posted on X that it takes misinformation seriously and will “prevent image creation on harmful topics.” The company noted that each AI-generated image carries a SynthID digital watermark that can be identified through its Gemini App or Lens in Search. Less than 24 hours after the feature’s debut, Google confirmed it would roll back the tool and work on stronger guardrails to enforce its content policies.