Particle.news

Google Pulls Google Earth AI Image Tool After Fabricated Satellite Scenes Circulate

The company says it will add stronger guardrails because doctored imagery risks undermining the trust journalists and investigators place in satellite photos.

Overview

  • Google added a “create image” feature using its Nano Banana 2 generator to Google Earth on July 30 and removed the tool roughly 24 hours later on July 31 while it builds stronger safety controls.
  • Open-source researchers and journalists showed the tool could produce photorealistic fakes of real places, including images of refugees at the U.S.-Mexico border, a nuclear plant in Iran, bomb craters in Gaza and staged crashes in Amsterdam.
  • Google said every generated image carried an invisible SynthID watermark and that images were sandboxed to the creator, but testers found prompts for harmful scenes were accepted and that watermarks can be lost or bypassed by screenshots and re-encodes.
  • Experts warned the episode could erode decades of trust in satellite and aerial imagery used by reporters, open-source investigators and humanitarians to verify events on the ground.
  • Google gave no timeline for restoring the feature and said it will only return after stronger guardrails are in place, leaving journalists and fact-checkers to rely on alternate satellite sources and multi-source verification for now.