Google has withdrawn a newly introduced AI-powered image generation feature for Google Earth less than a day after launching it, citing concerns over misuse and policy violations.

The experimental tool enabled users to create highly realistic images of real-world locations from text prompts using Google’s AI model and Google Earth’s mapping data. Soon after its release, users began generating fabricated scenes depicting wars, bomb damage, floods and other events that never occurred, raising fears that the technology could be used to spread misinformation.

Google said the feature was being used for legitimate purposes such as urban planning, architecture and education, but acknowledged that some generated content did not meet its safety standards. The company has now disabled the tool while it works on stronger safeguards before any potential relaunch.

Although every AI-generated image carried a visible watermark and was never published directly within the public Google Earth platform, experts warned that screenshots could easily circulate online without context, making fake location-based imagery appear authentic.

The rollback highlights the growing challenge facing AI companies as they balance rapid innovation with safeguards against the creation and spread of deceptive digital content.