Google Rolls Back AI Image Feature in Google Earth After Policy Violations

Google has reversed the rollout of its new AI image generation feature in Google Earth, withdrawing it less than two days after its introduction. The company stated that users had shared generated images which appeared to violate its policies, and that it is now working on implementing stronger safeguards.

The feature, launched on July 30, allowed users to create AI-generated images using Google Earth's satellite, aerial, and 3D imagery as a base, by inputting text prompts. Google had initially promoted it for uses such as recreating historical sites, generating educational infographics, and visualizing real estate projects.

However, researchers and experts had raised concerns about the potential for misuse, warning that users could create convincing fake satellite images by combining AI content with Google Earth's trusted mapping data. Tests by outlets like the BBC reportedly produced fabricated images, including a collapsed Eiffel Tower and Russian tanks in Kyiv.

Google had previously stated that all generated images included a SynthID digital watermark to indicate AI origin and that it prevented creation on harmful topics. Despite these measures, users reportedly bypassed some restrictions and shared images outside the Google Earth platform. The company has not specified a timeline for the feature's potential return.