Google has announced the suspension of its AI image generation feature in Google Earth, which was shut down less than 48 hours after launch due to risks of misinformation. Previously, Google integrated the AI image generation model Nano Banana2 into Google Earth, allowing users to generate fictional scenes over real satellite imagery using text prompts. BBC testing revealed that the feature could produce highly realistic but false scenarios such as “the Eiffel Tower collapsing,” “the pyramids being swallowed by sinkholes,” and “Russian tanks appearing in Kyiv,” making them easily mistaken for authentic satellite imagery. Security experts warned that synthetic maps could be exploited to spread misinformation and mislead the public and media during sensitive events such as wars or natural disasters. Google stated that all AI-generated images are embedded with invisible digital watermarks and can be verified using Gemini or Google Lens; however, BBC testing indicated that some verification mechanisms may be circumvented. This incident highlights that as generative AI enters high-trust information domains like mapping, preventing deepfakes and the spread of misinformation has become a critical security challenge.Author and source: AIBase
Google has announced a pause on its newly launched AI image generation feature in Google Earth. The feature, which was active for less than 48 hours, was temporarily disabled due to potential risks of misinformation, with the company stating it will enhance its security measures. Previously, Google integrated the Nano Banana2 image generation model into Google Earth, allowing users to generate fictional scenes via text prompts on top of real satellite imagery, aerial photos, and 3D maps.
According to BBC testing, this feature can generate false scenarios such as "the Eiffel Tower collapsing," "a massive sinkhole swallowing the Egyptian pyramids," and "Russian tanks appearing in Kyiv," and allows direct download and sharing. When combined with real geographic coordinates and Google Earth base maps, these AI-generated contents appear highly realistic and are easily mistaken for genuine satellite imagery.
Security experts warn that such highly realistic synthetic maps could be used to spread disinformation during sensitive events like wars or natural disasters, misleading the public, media, and even investigators, presenting new challenges for information verification.
Google states that all AI-generated images are embedded with invisible digital watermarks and support authenticity verification via Gemini or Google Lens. However, BBC testing found that some verification mechanisms can still be bypassed, and certain third-party detection tools cannot reliably identify AI-generated content.
This service suspension highlights that as generative AI increasingly enters high-trust information domains such as mapping and remote sensing, addressing security concerns—particularly how to enhance creative capabilities while preventing deepfakes and the spread of misinformation—has become a critical challenge in the deployment of AI products.