Google has rolled back its new Nano Banana AI image generation feature in Google Earth after users quickly used it to create realistic-looking fake scenes of disasters, conflict zones, explosions, and other misleading events.
The feature launched on July 30 as a creative tool for Google Earth web users. A day later, Google added an update saying it was pulling the capability while it works on stronger guardrails.
That fast reversal turned what was supposed to be a product showcase into another warning sign for AI tools built into trusted information platforms.
What Happened With Google Earth’s AI Image Generator?
Google introduced Nano Banana image generation in Google Earth as a way to “transform any place” using text prompts.
The company pitched the feature for uses like visualizing what Pompeii may have looked like in ancient times, turning an empty lot into a real estate concept, creating educational infographics, or imagining a future version of a familiar place.
The tool worked inside Google Earth on the web. Users could zoom into a real location, tap “create image,” then ask the system to generate a new visual based on Google Earth’s satellite, aerial, and 3D imagery.
That was the pitch.
The problem was what happened next.
Users and researchers began sharing examples showing how the same tool could generate fake crisis imagery on top of real-world places. Reports cited examples involving refugee camps, bomb craters, nuclear facilities, fatal crashes, and explosion scenes.
For Google Earth, that is not a small issue.
People use the platform as a reference point for what the world looks like. If that same interface can produce convincing fake scenes, the line between map, simulation, and visual evidence gets harder to explain.
Why Google Rolled It Back
Google said people had used the tool for useful geospatial purposes, but it had also seen screenshots of generated imagery that appeared to violate its policies.
The company said generated images did not appear in the main Google Earth experience for other users to see. It also said the images were watermarked as AI-generated.
Still, that did not stop the backlash.
The concern was not just that fake images could be created. The bigger issue was that they could be created inside a product many people already trust.
A fake image made in a random AI tool is one thing. A fake image framed like Google Earth imagery may carry more weight, especially when shared quickly on X, Facebook, Reddit, or messaging apps without context.
Why This Became A Misinformation Concern
Google Earth has long been used by journalists, researchers, investigators, and open-source intelligence communities to compare locations, check claims, and study events.
That trust is exactly why the Nano Banana rollout drew criticism.
If a user can generate a fake disaster scene over a real city block, then screenshot it and share it elsewhere, viewers may not immediately know it is artificial. Even if watermarks exist, most people scrolling through social media will not stop to verify every image.
This is the part Google now has to solve.
AI image generation is not new. But placing it directly inside a global mapping product changes the risk. It gives fake imagery a realistic geographic frame from the start.
What Is Nano Banana 2?
Nano Banana 2 is Google’s AI image generation model used for creating and editing visuals from text prompts.
Google has been adding Nano Banana-powered image tools across its products, including Gemini and other AI experiences. In Google Earth, the model was connected to real-world map imagery so users could build custom visuals from actual locations.
That made it useful for education, urban planning, design, and creative exploration.
It also made abuse more obvious.
A tool that can show what a community garden might look like on an empty lot can also be asked to show something that never happened, but looks close enough to confuse people.
What Happens Next?
As of this writing, Google has not announced a full relaunch date for Nano Banana image generation in Google Earth.
The company said it is working on stronger guardrails before bringing the feature back. That could mean stricter prompt blocking, clearer labels, stronger visible warnings, limited sharing options, improved verification tools, or restrictions on sensitive locations and crisis-related prompts.
The rollback also adds to a larger pattern across the tech industry: companies are racing to add generative AI features to major products, then adjusting them after real users expose risks that were easier to predict than to contain.
For now, the lesson is simple.
AI image generation may be useful inside creative tools. But inside a product people use to understand the real world, the margin for error is much smaller.
Suggested social embeds: News From Google on X | Henk van Ess on X | TechnoBaboy on Facebook
via: Search Engine Journal | Google Blog | Reuters | The Verge
