Google pauses Earth AI tool after fake satellite imagery sparks misinformation
The company withdrew a feature allowing AI-generated scenarios on real maps after it was used to create deceptive imagery.
Google has paused a new AI integration within Google Earth after the tool was used to generate deceptive imagery that mimicked authentic satellite data. The feature, which allowed users to overlay AI-generated scenarios onto real geographical maps, was withdrawn within 48 hours of its launch due to concerns over misinformation and policy violations.
The tool integrated an AI image generator known as Nano Banana 2 into the Google Earth platform. This allowed users to create realistic but fake visual scenarios on top of existing satellite imagery. According to the BBC, the feature was quickly exploited to produce misinformation, including a widely circulated fake image depicting a massive sinkhole swallowing the Great Pyramid of Giza.
The risk of AI hallucinations
This incident occurs as tech giants increasingly integrate generative AI into geospatial tools to enhance visualization and user experience. However, the deployment of such tools introduces the risk of "hallucinations"—where AI generates plausible but false data—or intentional manipulation. When these AI-generated elements are blended with real-world coordinates, the resulting imagery can be indistinguishable from authentic satellite intelligence to the untrained eye.
Undermining the 'ground truth'
The ability to easily spoof satellite data is particularly concerning because geospatial imagery is frequently treated as a "ground truth" in global affairs. Intelligence agencies, human rights monitors, and environmental organizations rely on this data to verify events on the ground, track war crimes, or monitor deforestation. The capacity to generate convincing fake imagery undermines the reliability of this evidence and creates a vector for geopolitical disinformation, where fabricated events could be presented as verified satellite sightings.
Future safeguards
Google has since withdrawn the feature to prevent further misuse. The company's rapid response highlights the tension between the push for generative AI innovation and the necessity of maintaining the integrity of mapping data. Observers will be watching how Google and other geospatial providers implement guardrails—such as digital watermarking or stricter content filters—before attempting to reintroduce AI-generative capabilities into their mapping ecosystems.