Google закрыла AI-редактор спутниковых снимков в Google Earth через сутки после запуска

Google launched a satellite image editing feature in Google Earth on July 30, 2026, powered by the AI model Nano Banana 2 — and by July 31, 2026, less than a day later, it was disabled. The reason: a researcher from Digital Digging clearly demonstrated how the tool allows users to add refugees near borders and traces of airstrikes next to hospitals onto real imagery.

What the AI Editor in Google Earth Could Do

The new feature embedded generative AI directly into the Google Earth interface: a user would select a region, enter a text prompt — and the Nano Banana 2 model would redraw the satellite image according to the description. Google positioned the tool as a means of visual exploration: the official demo example showed Pompeii reconstructed as it might have looked before the eruption in 79 AD.

Key launch parameters:

• Launch date — July 30, 2026
• Model — Nano Banana 2 (Google's generative AI model for images)
• How it works — text prompts applied over real satellite imagery
• Protection as stated by Google — a digital watermark on every result
• Built-in restrictions — blocking "the creation of images on harmful topics"

Users could add, remove, or modify objects on real maps — effectively editing images of Earth's surface in a public service with an audience of millions.

Why Google Shut Down the Feature the Next Day

Henk van Ess from the media outlet Digital Digging deliberately tested the limits of the tool and documented the results: using several text prompts, he was able to generate images showing groups of refugees at the Mexican border, a bomb crater near a hospital in the Gaza Strip, and a nuclear facility on Iranian territory. He described his experiments in an article titled "How to Plant a Nuclear Reactor in Iran," published the same day on the Digital Digging website.

Google's initial reaction to the criticism was restrained: the company cited watermarks and built-in content filters.

"Images created using

Nano Banana 2 in Google Earth include a digital watermark; the company prevents the creation of images on harmful topics," read the official response from Google's @NewsFromGoogle account on X.

However, on July 31, 2026, Google published a new statement and completely rolled back the feature. The tool had been live for less than 24 hours. The watermark proved to be insufficient protection: when shared on social media, it often goes unnoticed, and images resembling real satellite photography are perceived as documentary evidence.

What This Means

The Google Earth incident exposed a fundamental risk of AI-based geodata editing: even with filters and watermarks in place, generative AI is capable of creating convincing deepfakes of specific locations on Earth — and this tool is instantly used for information manipulation. This is particularly dangerous near state borders, in active conflict zones, and near civilian infrastructure facilities.

Frequently Asked Questions

What is Nano Banana 2?

Nano Banana 2 is Google's generative AI model for creating and editing images. It was the basis of the Google Earth feature, synthesizing modified versions of satellite images based on users' text prompts.

Why doesn't a watermark protect against deepfakes?

A digital watermark helps identify AI origin upon detailed inspection, but does not prevent distribution: on social media, users see the image, not the metadata. A convincing satellite image showing "refugees at the border" or "a crater near a hospital" can be shared thousands of times before anyone thinks to question its origin.