Google Pauses AI Image Generation in Earth After Misuse Concerns
Tool allowed users to create and download fabricated satellite imagery, prompting rapid rollback by the tech giant.
Google has temporarily suspended a new artificial intelligence feature in Google Earth that allowed users to generate and download edited satellite imagery, just one day after its launch. The company cited concerns over misuse after users quickly created and shared images depicting fabricated scenarios, including nuclear power plants in Iran and groups of refugees near the Mexican border.
The feature, integrated with Google's Nano Banana 2 image-generation model, enabled users to create reimaginings of real-world locations using text prompts. Users were able to generate downloadable false situations superimposed on base satellite, aerial, or 3D imagery. These capabilities were described as allowing users to create "concepts grounded in the real world" and "let imagination run wild."
Users quickly demonstrated the tool's potential for misinformation. OSINT expert Hank van Ess reported creating images of a nuclear plant in Iran and refugees at the Mexican border. He also generated an image of a bomb crater at a Gaza hospital. Van Ess noted that the tool did not initially refuse prompts related to harmful topics, nor did it provide warnings or offer alternative suggestions.
Google stated it observed users sharing screenshots of generated imagery that appeared to violate its policies. In response, the company announced Friday that it was rolling back the feature to "work on implementing stronger guardrails."
Google noted that all images created with the Nano Banana tool were watermarked and traceable to its Gemini AI model. The company also clarified that these generated images did not appear in the main Google Earth experience for other users to see.
This incident occurs amid growing concerns about the proliferation of AI-generated images and their potential to spread false information. Examples include AI-generated satellite images shared on social media platform X during Middle East conflicts, and a widely viewed fake video depicting Dubai's Burj Khalifa engulfed in flames.
The platform X has since warned creators against posting unlabeled AI-generated videos of armed conflict, threatening temporary suspension from its monetization program. Google had initially outlined five potential uses for the new software, including helping students visualize history, learning about landmarks, creating real estate plans, visualizing projects, and enabling creative imagination.
Google Earth, which holds over 280 billion images across 110 countries, is celebrating its 21st anniversary this year.
Sources
- www.dailymail.com - Google pulls new Earth AI tool after researchers show it can be used to put fake nuclear power plants in Iran and refugees near the Mexican border
- www.dailymail.com - Google pulls new Earth AI tool after researchers show it can be used to put fake nuclear power plants in Iran and refugees near the Mexican border