Contents
Google Earth’s Controversial AI Image Generator Pulled After 24 Hours Amid Disinformation Fears
A bold experiment by Google to integrate an AI image generator directly into the browser version of Google Earth raised immediate alarms among experts, leading to its swift withdrawal. Within hours of its launch, the tool—which allowed users to generate photorealistic alterations to satellite and 3D imagery—sparked fears that it could become a powerful engine for disinformation, threatening the credibility Google Earth has meticulously built over two decades.
The controversial feature was deactivated a mere 24 hours after its debut, following a wave of intense criticism.
The Rise and Fall of Nano Banana 2 in Google Earth
Innovative AI Meets Geographical Data
On July 30, 2026, Google introduced a groundbreaking feature in the browser-based version of Google Earth, powered by its Nano Banana 2 generative AI model. This innovation allowed users to craft new satellite images and three-dimensional visualizations from simple text prompts, directly leveraging Google’s vast geographical data. Initial marketing materials showcased its potential for positive applications: reconstructing ancient sites like Pompeii before the eruption of Mount Vesuvius, visualizing urban planning projects, or overlaying educational infographics onto famous landmarks such such as the Statue of Liberty.
The concept was simple yet powerful: users could select any location on the globe, click an “Create Image” button, and use a text command (prompt) to add new elements to the landscape. This could range from hypothetical buildings and disaster aftermaths to infrastructure projects.
The official Google Earth X (formerly Twitter) account announced the feature with enthusiasm:
Today we’re bringing AI image generation with Nano Banana to Google Earth, letting you virtually reimagine anywhere in the real world. 🍌 https://t.co/zsgvAnbYrl
For the first time, you can generate custom images using Google Earth’s satellite, aerial, and 3D imagery. 🌎 pic.twitter.com/yed45vLZOo
— Google Earth (@googleearth) July 30, 2026
The generation process relied on authentic satellite, aerial, and 3D data from Google Earth as its foundation. The AI model then seamlessly integrated the user-generated objects into the existing landscape, adapting them to match the lighting, colors, and perspective of the original imagery. This sophisticated integration made the final generated images remarkably difficult to distinguish from genuine satellite photographs.
Experts Sound the Alarm: A Threat to Credibility
Immediate Concerns from Open-Source Intelligence (OSINT) Professionals
Almost immediately after the feature’s release, information verification experts and investigative journalists raised serious concerns. They quickly warned that the tool could become a potent weapon for disinformation creators, enabling the production of highly plausible fake images anywhere in the world. This swift backlash highlighted the growing challenges posed by AI-generated fake news and disinformation.
Henk van Ess, a renowned OSINT analyst, famously demonstrated the tool’s potential for misuse. In his article, he described how, within seconds, he “placed” a refugee camp near the U.S.-Mexico border, “planted” a nuclear power plant in Iran, and “added” a fatal accident on a street in Amsterdam. His findings were shared widely:
Tonight 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
— henk van ess (@henkvaness) July 30, 2026
Experts emphasized that even if detection tools could eventually identify AI-generated forgeries, the damage could be substantial before any debunking occurred. The core issue was trust: Google had spent two decades building a reliable archive of satellite imagery, a resource often used to verify other visual information. By introducing a tool that could fabricate images, the company risked eroding this hard-won credibility in an instant.
Flawed Safeguards and Rapid Withdrawal
The Illusion of Security
In response to initial criticism, Google stated that every image generated by Nano Banana 2 would include a SynthID digital watermark, verifiable through the Gemini app or Google Lens. The company also claimed to block the creation of images on “harmful topics.”
However, practical tests quickly exposed the limitations of these safeguards. Henk van Ess’s prompts—involving refugee camps, Iranian nuclear power plants, or bomb attacks on hospitals—were not blocked, demonstrating significant weaknesses in content moderation.
Google’s official News account on X addressed the concerns:
We take misinformation seriously – every image created with Nano Banana in Google Earth includes the SynthID digital watermark, so if someone is unsure about an image, they can ask the Gemini app or use Lens in Search to see if the image was AI-generated. In addition, we prevent… pic.twitter.com/XF9r7TfTf2
— News from Google (@NewsFromGoogle) July 31, 2026
Furthermore, the SynthID watermark proved easily circumventable. When fake images are shared online, they are rarely distributed as original files with intact metadata. Instead, they appear as screenshots, screen recordings, or photos taken with a phone camera—formats in which the SynthID watermark becomes undetectable, making verification virtually impossible for the average user.
A 24-Hour Experiment Ends
Faced with an avalanche of criticism and undeniable demonstrations of the tool’s disinformation potential, Google announced the withdrawal of the feature on July 31, 2026—just 24 hours after its launch. The company posted on X, stating its decision to remove the functionality from Google Earth and to work on implementing stronger safeguards.
Re: our statement on Image Generation in Google Earth:
“We know that people uniquely trust Google Earth for a reliable view of the world. We’ve seen geospatial professionals using this feature for a range of useful purposes, however we’ve also seen people sharing screenshots of… pic.twitter.com/Yx7s7D1j2Z
— News from Google (@NewsFromGoogle) July 31, 2026
Google emphasized that none of the generated images appeared on the main Google Earth platform and all were marked as AI-generated within the experimental interface. However, experts quickly pointed out that this distinction was largely irrelevant once images were downloaded or screenshotted and shared on other platforms, where their origin and authenticity could not be easily verified. This incident underscores the ongoing challenges of integrating AI into mapping tools responsibly.
Verifying Satellite Imagery in the Age of AI Deepfakes
The case of Google Earth’s Nano Banana 2 highlights just how easily generative AI can be exploited to create widespread disinformation. In an era of increasing visual forgeries, relying on alternative and independent satellite data sources is more critical than ever for verifying information.
Key strategies for verification include:
- Cross-Referencing Multiple Sources: The simplest yet most effective method is to compare a suspicious image with at least two independent satellite sources. Free resources like Sentinel-2 (updated every 5-6 days), Landsat, Planet, Airbus, and Vantor offer distinct orbits and data collection times, providing diverse perspectives.
- Analyzing Visual Discrepancies: Pay close attention to subtle inconsistencies. In one notable instance involving a fabricated image of a military base in Bahrain, the discrepancy in the number of cars in a parking lot helped expose the forgery.
- Considering Context and Metadata: While easily removed, original metadata can sometimes offer clues. Always question images that lack context or source information.
Remaining vigilant and employing critical thinking are paramount when encountering satellite imagery online, especially as AI generation capabilities continue to advance.
The feature was withdrawn just 24 hours after its launch due to widespread criticism from information verification experts and OSINT analysts. They demonstrated how easily the tool could be used to create highly convincing fake satellite images, raising serious concerns about its potential for spreading disinformation and undermining Google Earth’s long-standing credibility.
To verify suspicious satellite images, compare them with at least two independent and reliable sources such as Sentinel-2, Landsat, Planet, Airbus, or Vantor. Look for inconsistencies in visual details, lighting, and objects. Be critical of images lacking clear context or source information, as AI-generated imagery can be highly realistic but often contains subtle discrepancies.
Source: BBC, DigitalDigging, 404media, Lifehacker, X, Futurism
Opening photo: Gemini
About Post Author
