Google had introduced an AI-powered image generation feature called Nano Banana 2 in Google Earth that enabled users to create manipulated imagery based on text prompts, drawing from the platform’s satellite, aerial, and 3D data. The feature was made available globally on the web version of Google Earth earlier this week, with the company positioning it as a tool for visualizing historical sites and real estate development projects.
Researchers at Digital Digging demonstrated the feature’s potential for creating misleading content by generating images depicting “refugees near the Mexican border” and a bomb crater adjacent to a hospital in Gaza. While some generated images were conspicuously artificial, such as a composite image merging the Sphinx with the Statue of Liberty, other outputs appeared sufficiently plausible to potentially deceive viewers and spread false information.
Google implemented SynthID digital watermarking on all images created through the feature and suggested users could verify AI-generated content using Gemini’s verification tools or Google Lens. The company also recommended comparing suspicious images against other satellite platforms, including Sentinel-2 or Landsat, and examining metadata such as capture dates and satellite information. However, researchers demonstrated that these safeguards were not entirely robust, successfully circumventing at least one AI detection system using an altered video from Google Earth.
Following the publication of concerns about the feature’s misinformation potential, Google announced it was rolling back the Nano Banana 2 capability on July 31st.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI