Alphabet’s Google said it was retracting an AI-powered image-generation feature from Google Earth a day after introducing it, citing instances in which users shared generated images that appeared to contravene the company’s policies. The capability had relied on Google’s Nano Banana 2 AI model to produce photorealistic images grounded in Google Earth’s satellite, aerial and 3D imagery based on user text prompts for any location on Earth.
In a message posted on X, Google acknowledged that geospatial professionals had used the tool for useful applications but that there were also examples of people posting screenshots of generated images that seem to violate the company’s policy standards. The company said it would pause the feature while it develops stronger guardrails.
Google did not specify what categories of images violated its policies. The company added that the generated imagery did not appear inside the principal Google Earth consumer experience and that those images were watermarked to indicate they were AI generated.
The move follows a broader pattern among major technology firms that have integrated AI features across products and then adjusted or restricted those capabilities in response to user behavior, misinformation concerns and policy issues. The company statement referenced this trend without naming additional specifics.
The Google Earth rollback echoes a recent action by Meta, which earlier this month discontinued its Muse Image feature. That tool had enabled users to generate images using public Instagram accounts and was pulled after widespread criticism over privacy concerns, including objections raised by a Hollywood union.
The Google announcement did not include further operational details, a timeline for restoring the feature, or specifics on the additional safeguards that will be implemented. Nor did it elaborate on the precise violations that prompted the pause. For now, Google says it will work on stronger protections to limit misuse before reinstating the functionality.
Contextual note: The paused capability produced AI-created visuals tied to precise real-world imagery, which critics said carried risks of creating misleading or falsified scenarios superimposed on authentic views of monuments and landmarks. Google emphasized that the images were watermarked and were not integrated into the main Earth experience.