Google Earth’s AI deepfake tool only lasted one day

Google shut down a Google Earth feature launched Thursday that let users edit satellite imagery with text prompts using Nano Banana 2. Researchers quickly demonstrated that the tool could generate misleading scenes, including refugees near the Mexican border and a bomb crater by a hospital in Gaza, despite Google saying outputs had watermarks and safety limits.

The episode highlights how AI-generated imagery can undermine trust in geospatial data when placed inside authoritative mapping products. Google’s rapid rollback suggests major platforms are still struggling to balance creative AI features with misinformation risks in high-stakes contexts like war, migration, and public safety.