Google launched an AI image generation feature in Google Earth using its "Nano Banana" model, allowing users to prompt-edit satellite imagery of real-world locations, only to roll it back less than 48 hours later after people began generating and sharing manipulated images that appeared to breach its policies. The episode matters because it undermined trust in a platform widely relied upon as an accurate visual record of the physical world, and drew criticism that Google should have anticipated the misuse given AI image tools' known potential for spreading misinformation.

The feature let users zoom into any location, click "Create Image," and enter a prompt to alter the satellite view, with Google promoting benign examples such as animal resorts and restored ancient ruins. Critics, including misinformation expert Henk van Ess, questioned the decision given Google's own acknowledgement that people "uniquely trust" Google Earth as a reliable source, drawing comparisons to Microsoft's 2016 Tay chatbot debacle as a cautionary precedent. Google said the images were watermarked as AI-generated, but has only paused the feature while developing stronger safeguards, rather than cancelling it outright.