Google Rolls Back Experimental Google Earth AI Feature After BBC Verify Raises Misinformation Concerns

Google spent months developing a new artificial intelligence capability for Google Earth. Less than two days after its late-July rollout, the feature was rolled back.

The decision was not driven by a software crash or technical bug—it followed warnings that realistic AI imagery could be mistaken for genuine satellite evidence, threatening the credibility of mapping data.

  

  

    

      

        

          Key Timeline & Fact Summary
        
      
    
    

      

        

          Event
        
        

          Feature rolled back less than 48 hours after release.
        
      
      

        

          Primary Cause
        
        

          Misinformation risks identified by BBC Verify.
        
      
      

        

          Scope Clarification
        
        

          Overlays were local and not on the public map.
        
      
      

        

          Current Status
        
        

          
            Feature remains paused while guardrails are built.
          
        
      
    
  

What Google Said in Response

In an official statement announcing the rollback, Google confirmed that screenshots violating its policies had begun circulating online, while emphasizing that standard Google Earth satellite views remain unchanged.

"We know that people uniquely trust Google Earth for a reliable view of the world," the company stated. "However, we've also seen people sharing screenshots of generated imagery that appear to violate our policies. So we're rolling back this feature in Google Earth while we work on implementing stronger guardrails."

Google further clarified that the experimental feature was designed purely as a local creation layer. AI-generated imagery was never automatically published to the shared global map viewed by other users; any synthetic scenes seen online were manual screenshots shared on external social platforms.

What BBC Verify Documented During Testing

Google initially introduced the capability as an experimental tool to visualize historical sites or conceptual overlays. However, during testing, BBC Verify documented how easily the generator responded to real-world crisis scenarios.

• Reported Model: BBC Verify reported that the feature used Google's Nano Banana 2 image model.

• Tested Scenarios: Reporters generated synthetic imagery depicting a non-existent nuclear facility in Iran, a refugee camp along the US–Mexico border, a damaged hospital in Gaza, and military vehicles in Kyiv.

• Verification Limits: While Google noted that its AI tools embed invisible SynthID digital watermarks, BBC Verify found that external AI detection tools failed in some cases to flag the manipulated imagery. In certain tests, Google's Gemini tool also misidentified synthetic satellite images as authentic.

Expert Analysis: Why Satellite Maps Demand Strict Guardrails

Satellite imagery carries a strong presumption of factual truth. When synthetic visuals are layered directly over real-world geographical coordinates, users are far more likely to assume the imagery represents verified location data.

Dutch open-source intelligence (OSINT) researcher Henk van Ess explained to BBC Verify that the feature allowed invented scenes to be "welded to genuine coordinates, drawn on genuine imagery."

"The forgery does not have to look convincing on its own," van Ess noted. "It inherits the credibility of the map it was born on."

Broader Industry Impact

The rapid rollback underscores a persistent challenge for tech platforms: how to offer flexible generative tools without compromising public trust in platforms built on objective geographical data. Google stated that the feature will remain paused while developers refine moderation systems and policy enforcement.

Reporting Sources & Transparency

• Primary Source: Official statement from Google LLC regarding policy violations, creation layers, and the feature rollback.

• Independent Investigation: BBC Verify reporting by Merlyn Thomas and Paul Brown.

• Expert Commentary: OSINT researcher Henk van Ess; AI-detection researcher Henry Ajder.

• Sourcing Transparency: Google's official statement and BBC Verify's independent investigation independently support different aspects of this reporting.