Google nixes its Earth AI feature one day after launch, amid criticism it would spread misinformation

Google has withdrawn an artificial-intelligence feature from Google Earth less than a day after introducing it, following rapid criticism that it could make deceptive map-based images easier to create and circulate. The short-lived tool paired the company’s satellite-imagery service with an image generator, allowing people to place invented visual material on top of views of actual locations. The episode has revived questions about what happens when generative systems are embedded in products that many users regard as factual reference tools.

The feature arrived on Thursday as part of an effort to make Google Earth a more imaginative environment. Google presented the option as a way for users to experiment creatively with geography rather than simply inspect the planet through maps and aerial imagery. It gave users access to Nano Banana 2, Google’s image-generation system, within the Earth experience. A written prompt could be used to produce a picture and position it within the mapping application’s real-world satellite views.

That combination was the focus of the backlash. Satellite maps carry an inherent air of documentation: they show recognizable streets, buildings, coastlines and terrain tied to a specific point on the globe. Adding an AI-created scene to that setting could make a fabricated image seem connected to a real place, even when the added event, object or condition never existed. Critics said the design could turn a familiar visual record into a vehicle for misleading claims.

The concern was not limited to sophisticated manipulation. Because the feature relied on prompts, it appeared to lower the barrier for making and sharing altered geographic imagery. Users would not necessarily need conventional image-editing skills to generate a persuasive-looking overlay. In practice, that could create a volume of synthetic location-based visuals that is difficult for ordinary viewers to assess, especially when such material travels beyond the Google Earth interface in screenshots or social posts.

Journalists and researchers were among those who highlighted the risk. Google Earth and related satellite imagery are often consulted as visual evidence when people seek to understand changes at a location or to corroborate reports. A BBC journalist responded to the launch with sarcasm, warning that a widely trusted source of visual information could be abused to distribute falsehoods. The reaction underscored that confidence in the underlying map can spill over to content placed on it.

Google’s decision to remove the feature came with unusual speed, reflecting how quickly the criticism took hold after Thursday’s rollout. The company had framed the capability around creative use, but the same flexibility that supported imaginative experimentation also created the possibility of visual claims with a real-world anchor. The story illustrates a recurring challenge for AI products: an intended entertainment or design function can take on a different significance when it is placed inside a service associated with verification.

There is also a distinction worth keeping in view. Generative imagery itself is not automatically misinformation, and creative treatments of maps can have legitimate uses in illustration, education or speculative design when their origin is plainly disclosed. The central objection was the prospect of synthetic additions appearing in a setting where audiences may expect an unaltered depiction of a real location. Clear signals about what is generated, and whether those signals survive when images are shared elsewhere, are likely to be central to future debates over similar tools.

For now, the removal means users no longer have the newly launched option to create Nano Banana 2 imagery inside Google Earth. It also leaves open broader questions for companies building generative AI into widely used information platforms. Product teams will face pressure to consider not only what a tool can make, but how it may be interpreted once it is detached from its original context. The next test will be whether future creative mapping features can preserve room for experimentation without weakening public confidence in visual geographic evidence.

Source: TechCrunch