Google rolled back an AI image-generation feature in Google Earth on July 31, one day after its launch, following widespread criticism that the tool could be used to fabricate and spread geospatial misinformation2,4.
The feature, which launched on July 30, integrated Google's Nano Banana 2 AI image generator into Google Earth, allowing users to generate custom images using the platform's satellite, aerial, and 3D imagery via text prompts. "For the first time, you can generate custom images using Google Earth's satellite, aerial, and 3D imagery alongside Nano Banana, which creates concepts grounded in the real world," wrote Bryan Horowitz, product manager for Google Earth, in a company blog post on July 30.
Critics moved quickly. Digital Digging's Henk van Ess intentionally generated images depicting "refugees near the Mexican border" and a bomb crater near a hospital in Gaza, demonstrating how easily the tool could produce fabricated scenes overlaid on authentic geographic imagery1. A BBC journalist posted sarcastically on Thursday: "There's no way that this new AI image generation feature on Google Earth, one of the most reliable sources of visual evidence for journalists and researchers, could possibly be abused to spread misinformation online".
Google initially defended the feature, pointing to built-in safeguards. The company said every image created with Nano Banana in Google Earth includes the SynthID digital watermark, and that users could verify whether an image was AI-generated by asking the Gemini app or using Lens in Search. Google also said it prevented "image creation on harmful topics".
Those defenses did not hold. By July 31, Google reversed course. "We've seen geospatial professionals using this feature for a range of useful purposes, however we've also seen people sharing screenshots of generated imagery that appear to violate our policies," the company said in a statement. "We're rolling back this feature in Google Earth while we work on implementing stronger guardrails".
The episode drew coverage from Bloomberg5, the Financial Times3,6, Business Insider, Ars Technica, The Verge, TechCrunch, and 404 Media. Ars Technica noted that while generative AI tools capable of modifying images of real buildings and cities already exist, Google's integration into Google Earth made it "even easier for anyone to create modified versions of authentic imagery depicting real locations".
ANALYSIS The rapid rollback highlights the tension between deploying generative AI capabilities inside trusted information platforms and the misinformation risks that follow. Google Earth has served as a reference tool for journalists, researchers, and open-source intelligence analysts; embedding an image generator directly into that environment lowered the barrier for producing fabricated geospatial imagery grounded in real-world locations. Google's reliance on post-hoc detection measures — SynthID watermarking and verification via Gemini or Lens — proved insufficient as a policy response once screenshots of generated imagery began circulating without those metadata layers intact. The one-day lifespan of the feature suggests that internal review processes did not fully anticipate the speed and scale at which adversarial use cases would surface publicly.