Google Pulls AI Image Generator From Google Earth After Deepfake Concerns

Google has temporarily removed a newly launched AI image-generation feature from Google Earth after it was found to be capable of creating convincing fake satellite images of real-world locations, raising concerns about misinformation and the misuse of geospatial data. 

The feature, powered by Google's Nano Banana 2 AI model, allowed users to generate photorealistic images by entering text prompts tied to any location on Earth. While Google introduced the tool to support creative visualization and geospatial applications, users quickly demonstrated its potential to fabricate scenes such as explosions, war zones, refugee camps, flood damage, and military installations. 

The rapid spread of AI-generated satellite imagery sparked criticism from open-source intelligence (OSINT) researchers, journalists and geospatial experts, who warned that such content could undermine public trust in satellite imagery, which is widely used to verify conflicts, natural disasters and humanitarian crises. Critics argued that even with digital watermarks, realistic fake imagery could still circulate widely on social media and be mistaken for authentic satellite photographs. 

Google confirmed it had paused the feature less than 24 hours after its rollout, saying it had observed users sharing AI-generated images that appeared to violate its policies. The company said it will strengthen safeguards before considering the feature's return. Google also noted that the generated images were clearly watermarked as AI-created and did not replace or alter the core Google Earth imagery available to users. 

The incident highlights the growing challenge facing technology companies as they integrate generative AI into widely trusted platforms. It also underscores increasing concerns over balancing innovation with safeguards to prevent misinformation and preserve the credibility of visual evidence online.