Google pulled its new AI tool that allowed people to edit satellite imagery within 24 hours of launch after some users demonstrated how the software could be misused.
The tech giant announced on Thursday it had integrated its Nano Banana 2 image-generation model into Google Earth, which allowed users to ‘type whatever you want to see’ and ‘virtually reimagine anywhere in the world’.
The AI tool allowed users to use text prompts to create downloadable false situations on top of a base layer of satellite imagery, aerial or 3D imagery.
However, users were able to quickly test the limits of the feature, described as allowing people to create ‘concepts grounded in the real world’ and let ‘imagination run wild’.
They shared images of fake nuclear plants in Iran, lines of refugees near the Mexican border, a fabricated bomb crater in Gaza and a car crash on a street in Amsterdam.
This prompted Google to quickly take the feature down on Friday, as it ‘worked on implementing stronger guardrails’.
In a statement on Friday, the company said it had seen ‘people sharing screenshots of generated imagery that appear to violate our policies’.
Google Street view holds more than 280 billion images across 110 countries, with Google Earth turning 21 this year.
Google pulled its new AI tool that allowed people to edit satellite imagery, including a fabricated bomb crater in Gaza
Hank van Ess, an OSINT expert and author of the Digital Digging newsletter, said he ‘planted a nuclear plant in Iran’ and ‘put refugees near Mexico border’ on Google Earth.
The tech expert also demonstrated how he was able to generate images of a bomb crater at a hospital in Gaza.
‘One clause in there is checkable tonight, so I checked it. Google prevents image creation on harmful topics,’ he wrote.
‘Refugees at a border. A nuclear plant in Iran. A fatal crash on an Amsterdam street. A hospital with a bomb crater in Gaza. Nothing was refused. No warning, no softened output, no suggestion to try a different prompt’.
The company noted that every image created with the Nano Banana tool included a watermark that can be traced back to its Gemini AI model.
This comes as concerns grow over the spread of AI-generated images on social media, with a greater ease for users to spread false information across the internet – which can be near impossible to contain.
While wars proliferated in the Middle East in March, AI-generated satellite images were shared on the social media platform X seemingly showing a successful Iranian drone strike on a US radar system in Qatar.
Meanwhile, a fake video purporting to show Dubai’s Burj Khalifa up in flames, while crowds of people fled from the building. It was viewed tens of millions of times.
Users were able to use the Nano Banana 2 image-generation model into Google Earth to create images, including a car crash on a street in Amsterdam
The proliferation of fake content prompted X to warn creators of a temporary suspension from its monetisation scheme should they post AI-generated videos of armed conflict without a label.
The programme rewards eligible users whose posts collect large numbers of views, likes, shares and comments with money.
In a blog post, Google shared five ways the new software could be used: Helping students visualise history; learn relevant information about particular areas or landmarks; create professional real estate plans; visualise projects; and letting ‘imagination run wild’ on favourite places.
A spokesperson for Google said: ‘We know that people uniquely trust Google Earth for a reliable view of the world.
‘We’ve seen geospatial professionals using this feature for a range of useful purposes, however we’ve also seen people sharing screenshots of generated imagery that appear to violate our policies.
‘So we’re rolling back this feature in Google Earth while we work on implementing stronger guardrails.
‘It’s important to note that generated images didn’t appear in the main Google Earth experience for others to see and were watermarked as AI generated.’.
