Google Rolls Back Google Earth AI Tool Over Misinformation Risks

Google rolled back a new AI image-generation feature in Google Earth on July 31, 2026, less than 48 hours after its July 30 launch. The tool, powered by the Nano Banana 2 model, allowed users to superimpose photorealistic AI visualizations onto actual satellite, aerial, and 3D maps using text prompts. While marketed for educational purposes and real estate planning, the feature was quickly exploited to create deceptive imagery of nuclear plants in Iran, bomb craters in Gaza and Russia, and refugees at the US-Mexico border.

Independent researchers and organizations, including BBC Verify and Bellingcat, demonstrated that the tool's safeguards were insufficient. Tests showed that users could bypass prompt filters and that SynthID digital watermarks were easily neutralized via screenshots. Experts warned that such forgeries inherit the perceived credibility of Google Earth's trusted coordinates, potentially undermining the reliability of geospatial data used by courts and war crimes investigators.

Google initially defended the feature by noting that generated images did not appear in the main public experience. However, following a surge of policy-violating content and intense backlash from groups like the Centre for Information Resilience and GeoConfirmed, the company paused the tool. Google stated it is implementing stronger guardrails before any potential return to the platform.