Google Earth Integrates AI Image Tool, Prompting Disinformation Warnings

Google integrated its Nano Banana 2 image generation system into Google Earth, allowing users to modify any global location with artificial intelligence. While demonstrated with whimsical concepts like dragons over Paris, disinformation researchers warn the tool compromises satellite verification by enabling malicious actors to fabricate highly realistic geographic imagery.

How Nano Banana 2 Transforms Google Earth

Google rolled out the update featuring its Nano Banana 2 image generation engine directly inside Google Earth. A new créer une image icon allows users to prompt environmental changes using the platform’s underlying 3D cartography, aerial data, and satellite archives.

The company promotes the feature as a creative utility enabling users that visualize history, create real estate projects, and more. Testers demonstrated the system by prompting climate control units onto Parisian rooftops to simulate future urban conditions, or adding fantastical elements like flying creatures to the skyline. The generator applies realistic shadows across roofs and beneath added objects, demonstrating technical polish alongside its imaginative capabilities.

Beyond scenic modifications, Google suggests using the generator to build educational infographics about monuments. Tests involving the Sacré-Cœur basilica in Montmartre produced detailed architectural drawings, though they also exposed inherent artificial intelligence flaws, such as generating factual errors regarding the precise square footage of the structure’s mosaic work.

Google removes AI tool from Google Earth after deepfake fears

Security Experts Warn of Compromised Verification Tools

Security researchers and disinformation experts immediately condemned the rollout, arguing that embedding generative tools inside a trusted cartographic repository undermines open-source investigation.

Google a passé vingt ans à construire la référence à laquelle le monde entier se fie. Aujourd’hui, elle ajoute un bouton qui invente des choses. Henk van Ess, digital investigation expert, via Substack

Investigators point out that manipulated geographic content requires minimal sophistication to deceive audiences because it inherits the trusted reputation of the underlying map database. Henk van Ess noted that the falsification does not even need to be convincing on its own because it inherits the credibility of the map on which it is founded.

In practical trials, journalists and analysts easily fabricated potentially dangerous imagery using the updated platform, generating convincing fake satellite views ranging from Iranian nuclear facilities and bomb craters in Russia to militant training camps in Syria. Similar synthetic imagery has driven real-world fallout in past conflicts, including a 2023 fake image of a Pentagon explosion that briefly disrupted financial markets, and altered satellite frames during early hostilities between the United States and Iran that amassed millions of views online.

Google Earth Integrates AI Image Generator via Nano Banana 2

Je ne saurais trop insister sur le caractère irresponsable de cette décision. Il a fallu des décennies pour bâtir la confiance dans les images satellites, et celle-ci pourrait être irrémédiablement détruite du jour au lendemain. Ross Burley, executive director of the Centre for Information Resilience, via AFP

Additional pushback came from analytical groups including the American Enterprise Institute’s satellite imagery specialist Brady Africk, who warned that the update facilitates the creation of convincing fake satellite images that are likely to spread rapidly online and mislead the public. GeoConfirmed, a collaborative conflict-monitoring network, reported observing users immediately leveraging the utility to establish fake military positions and mislabel civilian buildings as schools to manipulate media narratives.

Why Google Search Queries Are Getting Longer

Google Response and Safeguards

In response to widespread criticism, Google stated that it takes disinformation seriously. The technology giant asserts that all images produced via the feature contain invisible digital markers that users can detect using the Gemini chatbot.

Photo: Frandroid

Furthermore, the corporation maintains that automated filters prevent the creation of images on dangerous subjects and enforces an AI usage policy that explicitly forbids misleading presentations of fact or deliberate deception. Despite these internal guardrails, verification groups and investigative directors continue to urge Google to suspend the feature entirely to protect public trust in mapping infrastructure.