Summary
- Google discontinued an AI-driven image-generation feature from Google Earth on July 31, just one day post-launch, citing violations of its policies by users sharing generated images.
- Investigative reports revealed that the Nano Banana tool could easily create fictitious events, such as a blast crater in Los Angeles and a flooded U.S. Capitol, raising alarms over potential misinformation.
- Despite Google's assertion that images would carry a SynthID watermark to denote AI generation, critics remained unconvinced. The company stated it would only restore the feature after implementing stronger safeguards, with no specific timeline provided.
Google has retracted a recently launched AI feature from Google Earth just a day after it became available, responding to immediate concerns raised by journalists and open-source researchers about the tool's potential to generate misleading satellite images.
The feature was introduced on July 30, allowing users to navigate to any spot on Google Earth's web interface, click "create image," and generate scenes through text prompts using the Nano Banana model. However, by July 31, it was no longer accessible.
In a statement shared on X, Google emphasized that users โuniquely trust Google Earth for a reliable view of the world,โ noting that while geospatial professionals found the tool beneficial, others misused it to create images that breached guidelines. The company announced plans to retract the feature while enhancing its safeguards.
โTonight I typed just one sentence into Google Earth and put refugees near the Mexican border. Then I planted a nuclear plant in Iran. What on earth is Google doing? Check my latest post here: https://t.co/N0h6CVECqS pic.twitter.com/WSjdcEXWJx
โ ๐๐๐๐ ๐๐๐ ๐๐๐ (@henkvaness) July 30, 2026
Concerns revolved around the tool's ability to fabricate entirely false events. The tech site 404 Media showcased how it could generate a blast crater in Los Angeles and even add protesters outside Google's Mountain View headquarters. NPR also created images depicting Iran's Kharg Island in flames and a submerged U.S. Capitol, both of which would be significant news stories if they were authentic. Open-source researcher Henk van Ess reported trying various prompts, including one about refugees at the Mexican border and another about a nuclear facility in Iran, with no prompts being rejected.
Regarding our statement on Image Generation in Google Earth:
โWe know that people uniquely trust Google Earth for a reliable view of the world. Weโve seen geospatial professionals using this feature for a range of useful purposes, however weโve also seen people sharing screenshots ofโฆโ News from Google (@NewsFromGoogle) July 31, 2026
Satellite imagery has traditionally been a reliable source for validating breaking news and documenting crises, largely because it has been challenging to fabricate. Researcher Jake Godin from Bellingcat warned that the ease of generating these images could facilitate the rapid spread of misinformation, with the potential for governments to dismiss genuine images as forgeries.
Initially, Google downplayed the risks, asserting that each image would carry a SynthID watermark, marking it as AI-generated in applications like Gemini, and that it would block the creation of images on harmful subjects. Critics, however, argued that most individuals do not verify images before sharing them.
The company stated it would only reinstate the image generation feature in Google Earth after implementing more robust protections, without providing a timeline for this process.