Google abruptly removed its new AI image generation feature from Google Earth on Friday, July 31, 2026, less than 24 hours after its launch. The feature, rolled out on Thursday, allowed users to type prompts and have the AI model Nano Banana 2 superimpose fabricated scenes onto real satellite, aerial, and 3D map imagery. A product manager had touted it as a way to “visualize history, create real estate plans and more.”
However, backlash erupted almost immediately. Open-source investigator Henk van Ess criticized the tool, saying Google spent two decades building a trustworthy reference only to add a “button that makes things up.” Journalists and researchers from BBC Verify, Bellingcat, and others quickly demonstrated how easily the feature could be abused. They generated images of a collapsed Eiffel Tower, Russian tanks in Kyiv, a bomb crater near a Gaza hospital, and Iran’s Kharg Island in flames, among many other deceptive scenes.
Google confirmed the rollback by Friday, stating that while some geospatial professionals found the tool useful, the company observed users sharing generated imagery that violated its policies. The company said it is working on implementing stronger guardrails. Google had attempted safeguards, including a SynthID digital watermark and content filters, but these proved easily bypassed—BBC Verify even tricked the Gemini app into authenticating a fake image.
The incident underscores the growing tension between generative AI innovation and the risk of misinformation. For a platform long considered a reliable source of visual evidence, the one-click creation of convincing fakes threatened to erode trust in satellite imagery, as highlighted by analysts like Jake Godin and Bill Greer. The rapid withdrawal serves as a case study in the challenges of deploying AI in sensitive contexts, and it remains unclear if the feature will return.
