Google shut down a Google Earth feature that let anyone generate fake AI imagery over real satellite maps, after researchers showed how easily it produced convincing misinformation.
Google launched an AI image-generation feature inside Google Earth that let users edit satellite, aerial, and 3D imagery with text prompts, then killed it within 24 hours after Digital Digging researcher Henk van Ess and others demonstrated how easily it produced realistic, false depictions of real locations. The Verge reports the backlash was immediate and specific to the tool's potential for geographic misinformation.
This is one of the fastest product reversals from a major AI lab this year, suggesting internal review processes didn't catch what outside testers found in hours.
Shipping first and rolling back after public backlash is now a visible pattern at Google, and it's an expensive one: reputational cost, wasted engineering cycles, and a green light for competitors to point to it as evidence of rushed AI safety review. Any product team building generative tools on top of authoritative data sources like maps needs red-teaming before launch, not after.
The daily signal, curated. Get it in your inbox.
Subscribe on LinkedIn →