Nuclear plants, Gaza craters: Google pulls AI tool from Google Earth in 24 hours

Google has swiftly removed a feature from its popular Google Earth platform following just 24 hours of its debut. The AI-generated satellite imagery tool, introduced on July 30, allowed users to create visualizations on top of satellite images with a simple click. However, researchers and open-source intelligence experts raised concerns about potential risks of disinformation that the feature posed.

Google's decision to rollback the feature came as a response to these concerns, with the company stating it was designed to align with users' imaginations. The tool, powered by Google's Nano Banana 2 image-generation technology, enabled users to generate images based on Google Earth satellite, aerial, and 3D-mapping data. Despite being watermarked as AI-generated and placed within Google Earth without being visible to others, the generated images were found to violate policies.

Experts warned that the feature could potentially lead to the spread of convincing fake satellite imagery, which could mislead the public and erode trust in satellite imagery data. Google countered these concerns by emphasizing the presence of SynthID digital watermarks in the generated images, which can be verified using the Gemini chatbot or Lens in Search.