Google Earth's new AI image generation function didn't survive a day after users deepfaked disasters

Aug 01, 2026 03:55 AM - 1 hour ago 1

Godzilla fires a bluish power beam complete a burning metropolis while subject jets alert nearby.

Google pulled a caller AI image-editing characteristic from Google Earth little than 24 hours aft launch,

The institution launched the usability connected Thursday, allowing Gemini's Nano Banana 2 image exemplary to entree Google Earth and toggle shape locations pinch matter prompts.

The function, however, quickly sewage retired of hand. AI interrogator Henk van Ess, for example, created highly realistic deepfakes, including clone exile camps on the US-Mexico separator and a atomic works successful Iran.

Other users utilized the instrumentality to create AI images of a nonexistent flood and a 9/11-style attack.

By Friday, Google had reversed course.

"We've seen geospatial professionals utilizing this characteristic for a scope of useful purposes, nevertheless we've besides seen group sharing screenshots of generated imagery that look to break our policies," a Google spokesperson said successful a connection posted connected X. "So we're rolling backmost this characteristic successful Google Earth while we activity connected implementing stronger guardrails."

Google added that the AI-generated images were watermarked and ne'er appeared successful the nationalist Google Earth experience.

Google is not the only institution that has lately rolled retired AI features only to quickly region them. Meta launched Muse Image connected July 7 pinch a characteristic that allowed users to make AI images utilizing images disposable connected each adult-owned nationalist Instagram accounts that had not changed their settings to opt out. Meta removed the characteristic successful little than a week and said it had "missed the mark."