Google pulls Google Earth AI image editor one day after launch
Google shut down a Google Earth feature that let users edit satellite images with AI text prompts, just one day after it launched on a Thursday. The tool, which ran on Google's Nano Banana 2 image model, effectively let anyone create AI deepfakes of real-world locations. Henk van Ess of Digital Digging tested the limits by intentionally generating images that added refugees near the Mexican border and a bomb crater next to a hospital in Gaza. When van Ess raised this, Google's initial response said images generated in Google Earth carried a digital watermark and that the company blocked "image creation on harmful topics." One day later, Google reversed course. In a new statement, the company said: "We know that people uniquely trust Google Earth for a reliable view of the world. We've seen geospatial professionals using this feature for a range of useful purposes, however we've also seen people sharing screenshots of generated imagery that appear to violate our policies. So we're rolling back this feature in Google Earth while we work on implementing stronger guardrails. It's important to note that generated images didn't appear in the main Google Earth experience for others to see and were watermarked as AI generated." Van Ess pushed back on the watermark defense: he said he was able to fool Hive's AI detection tool with a video generated in Google Earth, showing that the generated media could spread and pass as real despite the watermark. He also said the tool's guardrails did not work as described during his testing: "Nothing was refused, nothing was softened, and nothing suggested I try a different prompt." The source does not say how many users tried the feature, how many policy-violating images were generated before the rollback, what the promised stronger guardrails will consist of, or whether the feature will return once they are built.
Key facts
- Google launched an AI image-editing feature in Google Earth on a Thursday and shut it down one day later.
- The feature ran on Nano Banana 2 and let users alter real satellite imagery with text prompts.
- Henk van Ess of Digital Digging generated fabricated images, including refugees near the Mexican border and a bomb crater by a Gaza hospital, to test the tool's limits.
- Google said generated images carried a digital watermark and did not appear in the main Google Earth view for other users.
- Van Ess said he fooled Hive's AI detection tool with a Google Earth generated video, and that the tool refused none of his prompts during testing.
Why it matters
A major mapping platform that people trust for a reliable view of the real world briefly shipped a tool that let anyone fabricate convincing scenes at real locations, from refugees at a border to a bomb crater by a hospital. Google's own statement acknowledges the core tension: the same trust that makes Google Earth useful is what makes AI-altered imagery from it dangerous.
Who it affects
Google Earth users who had access to the AI editing feature, geospatial professionals who Google says were using it for legitimate purposes, and anyone who might encounter generated Google Earth imagery or video circulating online, since the source shows such content can pass as authentic.
How to use it
The feature has been rolled back, so it is not currently available. Google says it is working on stronger guardrails before any return; the source does not say when or whether that will happen.
How solid is it
The account rests on Google's two public statements and on Henk van Ess's own testing, including his claim that he defeated Hive's AI detection tool with a generated video. The source does not independently verify Hive's detection failure or quantify how widely the harmful images spread before the rollback.
Risks and caveats
Google's watermarking and content-policy defenses did not hold up in van Ess's testing: he reported no refusals or softened responses to his prompts, and he says he beat an AI detector with generated video, undercutting the claim that watermarks are enough to contain the risk. The source does not disclose how many users accessed the feature, how many violating images were made, or what the new guardrails will actually consist of.
“We know that people uniquely trust Google Earth for a reliable view of the world. We've seen geospatial professionals using this feature for a range of useful purposes, however we've also seen people sharing screenshots of generated imagery that appear to violate our policies. So we're rolling back this feature in Google Earth while we work on implementing stronger guardrails.”
— Google, in its rollback statement