Google has rolled back a new Google Earth feature just one day after introducing it, after critics warned that the tool could make fabricated geographic imagery easier to create and share.
The feature brought Nano Banana 2, Google's AI image generator, into Google Earth. Users could prompt the system to create artificial images and place them inside the satellite imagery mapping app, with Google presenting the idea as a way to get creative with geography.
What Google launched
The removed Google Earth AI feature was prompt-based. According to the source article, it allowed users to superimpose almost any generated image over real maps inside Google Earth.
That combination made the feature unusually sensitive. Google Earth is not just another visual canvas. It is widely treated as a place to inspect geography, compare locations, and view satellite imagery that appears connected to real places.
By adding an AI image generator directly into that environment, Google blurred a line that many users rely on: the difference between a map-based view of the world and an invented scene placed on top of one.
Why critics reacted quickly
The main concern was misinformation. If fabricated images can be placed over real geographic imagery, screenshots can appear more credible than ordinary AI images because they are framed inside a familiar mapping product.
Critics argued that the tool could produce what the source described as geospatial slop: generated visual material tied to real places but not representing reality. The risk is not simply that an image is fake. The risk is that the map context may make the fake image seem more trustworthy.
A BBC journalist captured that concern sarcastically in a post on Thursday:
"There's no way that this new AI image generation feature on Google Earth, one of the most reliable sources of visual evidence for journalists and researchers, could possibly be abused to spread misinformation online,"
That reaction points to the core problem for Google. Google Earth carries a reputation as a visual reference point for journalists and researchers. A feature that makes it easier to create convincing but fabricated map imagery can weaken confidence in screenshots and shared visuals, even when the underlying product remains useful.
Google's rollback
Only a day after the feature was released, Google scrapped it. The company said some geospatial professionals had used the tool in helpful ways, but it also saw examples that raised policy concerns.
Google said:
"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,"
The company added:
"We’re rolling back this feature in Google Earth while we work on implementing stronger guardrails,”
The wording matters. Google did not describe the idea as useless. It pointed to useful purposes among geospatial professionals. But it also acknowledged that generated imagery was already being shared in ways that appeared to violate its policies.
That is the tension many AI tools now face. A feature can have legitimate creative or professional uses and still create obvious opportunities for abuse when the output looks connected to real-world evidence.
The bigger AI image problem
The criticism of the Google Earth feature is understandable, but the source article also notes a wider reality: in the age of AI, most imagery on the web can be manipulated easily.
Creating semi-credible disinformation no longer requires advanced image-editing skill. The source makes the point plainly: a user does not need to be a Photoshop expert anymore. Access to an AI image generator can be enough.
That broader context does not remove Google's responsibility for product design. If anything, it raises the stakes. When image generation is placed inside a product associated with real locations and visual evidence, the surrounding interface can give the output a credibility boost.
The Google Earth case shows why guardrails are not a secondary detail for AI image tools. They shape how generated content can be understood, shared, and misused.
What this episode shows
Google's decision to roll back the feature so quickly suggests that the company saw the risk as immediate. The issue was not just theoretical criticism. Google said people were sharing screenshots of generated imagery that appeared to violate its policies.
For users, the episode is a reminder to treat map screenshots and location-based visuals with care when AI tools are involved. For AI companies, it shows that embedding generation inside trusted information products creates different risks than offering a standalone creative tool.
The central lesson is straightforward: when fabricated images can sit on top of real maps, context becomes part of the illusion. Stronger guardrails are not only about preventing bad prompts. They are about preserving the difference between creative geography and visual misinformation.