Google Pulls Google Earth AI Tool Less Than 24 Hours After Launch Following Fake Imagery Backlash
Google rolled back a newly launched AI image generation feature in Google Earth on Friday, July 31, less than 24 hours after unveiling it, following backlash from researchers and open-source intelligence experts who warned the tool could be used to create convincing fake satellite imagery.
What the Tool Did
Google introduced the feature Thursday, July 30, integrating its Nano Banana 2 image-generation technology directly into Google Earth. According to Forbes, the tool let users zoom into any location on the web version of Google Earth, tap “create image,” and generate photorealistic or illustrated scenes based on a text prompt — for example, visualizing what Pompeii may have looked like before the eruption of Mount Vesuvius, or turning a landmark into an illustrated infographic populated with data pulled from Google’s Gemini AI system.
In the company’s initial announcement, product manager Bryan Horowitz described the feature as designed to work with users’ imaginations, inviting people to type in “whatever you want to see” after zooming into a location.
How It Went Wrong
The rollout didn’t unfold as planned. According to Forbes, investigative journalist Henk van Ess, along with newsrooms including NPR and Futurism, demonstrated that the tool could be used to generate potentially misleading content — including images depicting refugees near the Mexican border and a fabricated nuclear power plant in Iran that didn’t actually exist.
TechXplore’s summary of independent testing found the tool could produce false depictions of a range of sensitive scenarios, including explosions, military sites, and floods. While the generated images did include SynthID watermarks identifying them as AI-generated, experts warned that the tool still risked undermining trust in geospatial verification more broadly — particularly because screenshots of the generated images could easily be cropped or shared without those watermarks remaining visible or noticed.
Why Satellite Imagery Verification Matters So Much
The concern wasn’t abstract for researchers who rely on satellite imagery professionally. Jake Godin, a senior researcher with Bellingcat, the online group known for visual forensic investigations, explained to NPR why satellite imagery has historically been considered one of the more reliable tools for verifying real-world events: it’s controlled by companies or institutions, and its source — a camera positioned hundreds of miles above the Earth’s surface — has been difficult to convincingly fake, unlike ground-level photos or video.
Godin noted that fake satellite imagery has already surfaced in limited cases previously, including early in the U.S.-Israel conflict with Iran, according to NPR’s reporting. A tool that made generating convincing fake satellite imagery significantly easier, in other words, threatened to erode one of the few remaining categories of visual evidence that had proven relatively resistant to fabrication.
Google’s Response
Google acknowledged the concerns directly in a statement provided to AFP and cited by TechXplore. “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,” a company spokesperson said. “So we’re rolling back this feature in Google Earth while we work on implementing stronger guardrails.”
The company emphasized the platform’s reputation for reliability in the same statement, adding, “We know that people uniquely trust Google Earth for a reliable view of the world.” Google also noted, in a follow-up statement cited by Forbes, that the generated images hadn’t appeared within the main Google Earth experience where other users could see them, and that all outputs were watermarked as AI-generated.
What Happens Next
Google has not specified exactly what improved safeguards it plans to implement before potentially reintroducing the feature. Forbes noted there’s a meaningful difference in real-world impact between the tool’s more benign use cases — like educational infographics about historical sites or famous landmarks — and its potential to generate convincing images of fabricated geopolitical events, meaning Google will likely need to draw much narrower boundaries around what the tool can generate if it returns.
FAQ
Is the Google Earth AI image tool still available? No. Google rolled the feature back on July 31, less than 24 hours after its July 30 launch, and has not announced a timeline for reintroducing it.
What kinds of fake images did the tool generate? Testing by journalists and researchers produced fabricated imagery including a fictitious nuclear power plant in Iran, refugees near the Mexican border, and other depictions of explosions, military sites, and floods, according to reporting from Forbes and TechXplore.
Were the AI-generated images labeled as fake? Yes. Google said all generated images included SynthID watermarks identifying them as AI-generated, though critics noted screenshots could be shared without those watermarks being noticed.
Why is fake satellite imagery considered particularly concerning? Researchers like Bellingcat’s Jake Godin note that satellite imagery has historically been one of the more difficult types of visual evidence to convincingly fake, making it a trusted tool for verifying real-world events. A tool that makes generating fake satellite-style imagery easy threatens that reliability.
Did the AI-generated images appear in the main Google Earth product for other users to see? No, according to Google’s statement. The company said the generated images did not appear in the main Google Earth experience visible to other users.
Conclusion
Google’s rapid rollback — pulling the feature within a day of launch — reflects how quickly concerns about AI-generated satellite imagery escalated once researchers demonstrated its potential for creating convincing fabricated geopolitical content. With Google saying it’s working on stronger guardrails before any potential relaunch, the episode adds to a broader pattern of AI companies moving fast on new generative capabilities and then having to walk them back once real-world misuse becomes apparent.
Sources: Bloomberg, NPR, Forbes, TechXplore (reporting dated July 31-August 1, 2026).