Google pulls AI image tool from Google Earth after deepfake misuse

Feature launch and swift reversal

Google introduced an AI image-generation feature in Google Earth on July 30, 2026, allowing users to create photorealistic images grounded in the platform's satellite, aerial and 3D imagery through text prompts. The tool was powered by Google's Nano Banana 2 AI model, which had previously gone viral. Less than a day later, on July 31, Google announced it was pulling the capability back while it worked on stronger guardrails.

Misuse and disinformation concerns

Users quickly exploited the tool to fabricate scenes including aerial images of destruction at Singapore's Changi Airport, a fire at Gardens by the Bay, a collapsed bridge near Marina Bay Sands, a plane appearing to crash into a Manhattan skyscraper, a nuclear facility in Iran and a bombing in Moscow. Screenshots circulated on X showing realistic satellite-style images of false terror attacks, industrial accidents and violence. Open-source intelligence investigators, journalists and AI experts warned the feature made disinformation easier to produce and spread; one early critique came from investigator Henk van Ess, who argued the tool would cause more harm than good. CNA's own tests found that some commercial AI-detection tools rated the fabricated images as authentic, while Google's Gemini app correctly identified them as AI-generated.

Google's stated response

Google said on X on July 31 that it had "seen geospatial professionals using this feature for a range of useful purposes" but had also observed "people sharing screenshots of generated imagery that appear to violate our policies." The company said it was "rolling back this feature in Google Earth while we work on implementing stronger guardrails" and noted the images did not appear in the main Google Earth experience and were watermarked through its SynthID technology. Google did not specify which categories of images violated its policies, and CNA noted that the public Google Earth map itself was not altered by user activity.

Broader context and industry precedent

The episode fits a wider pattern of AI image tools being restricted after misuse. Reporting in The Seattle Times noted that Google previously temporarily restricted Gemini image creation after accusations of bias, and that an AI-generated fake video created with Google's technology went viral last year. Experts cited across the coverage cautioned that satellite imagery has long served as a standard reference for verifying locations and debunking falsehoods in conflict and disaster reporting, raising the stakes of placing generative AI directly inside Google Earth. The Hindu's report referenced similar concerns about propaganda and false news tied to conflicts and environmental disasters, while The Star noted that Meta had earlier in the month discontinued its Muse Image AI feature after privacy criticism.