Google pulled a generative AI feature in Google Earth that let anyone superimpose fake imagery onto real satellite maps, after critics showed it could fabricate convincing deepfakes of real locations.
Google Earth's new AI editing tool allowed users to alter satellite, aerial, and 3D imagery with text prompts. Within a day, researchers including Digital Digging's Henk van Ess demonstrated the tool could generate misleading, reality-warping images tied to real coordinates, prompting immediate backlash.
Google rolled the feature back rather than adding guardrails or watermarking, opting for a full retreat over a patch.
Shipping generative tools tied to authoritative data sources like maps carries reputational risk that speed-to-market doesn't offset. Product teams should pressure-test misinformation vectors before launch, not rely on public backlash to catch them.
The daily signal, curated. Get it in your inbox.
Subscribe on LinkedIn →