MOUNTAIN VIEW — Alphabet’s Google has announced the rollback of an AI image-generation feature in Google Earth just one day after its launch, following user-shared creations that reportedly violated company policies. Powered by Google’s Nano Banana 2 AI model, the tool allowed users to generate photorealistic images anchored in Google Earth’s satellite, aerial, and 3D data by typing text prompts for locations worldwide. While geospatial professionals utilized the feature for constructive tasks, critics quickly raised alarms over its potential to generate convincing misinformation by superimposing fake scenarios onto real-world landmarks and monuments.
Google stated on X that it is pausing the capability to implement stronger guardrails, noting that the AI-generated images carried watermarks and did not appear in the primary Google Earth experience. This swift retraction underscores the ongoing challenges tech companies face when integrating generative AI across mainstream products, frequently forcing them to restrict or disable features in response to safety violations. The move mirrors recent industry friction, such as Meta’s decision earlier in the month to discontinue its Muse Image feature following widespread criticism over privacy and public data usage.
