Google Earth removed an AI image generation feature just one day after launching it. The tool allowed users to edit satellite imagery with text prompts, effectively letting people create deepfakes of real locations. Testers quickly demonstrated the risk by adding structures like refugee camps and bomb craters to sensitive areas. Google initially claimed the system included digital watermarks and blocked harmful topics, but the company deleted the function entirely within hours.
This incident highlights the difficulty of controlling generative AI when it interacts directly with geospatial data. The core issue is that synthetic imagery can be mistaken for factual records, potentially misleading researchers or influencing public perception of conflicts. Removing the tool suggests Google prioritised preventing misinformation over offering experimental editing capabilities.
* The feature was active for only 24 hours before deletion.
* Users could generate images of real-world locations using text.
* Google removed the function to stop the creation of harmful deepfakes.
