GOOGLE has rolled back its new Artificial Intelligence (AI) tool that allowed people to create fake scenarios on top of real-life satellite imagery after it faced criticism for potentially spreading misinformation.
Less than 48 hours after its release, the tech giant announced it was “rolling back” the feature while it implemented “stronger guardrails” due to concerns about users violating its policies.
In an initial statement announcing the release of the new “Nano Banana” update, Google Earth said the tool would allow users to “virtually reimagine anywhere in the world,” showing a user using AI to recreate the Colosseum before the fall of the Roman Empire.
It actively called on people to create fake versions of reality, setting challenges to create “pet paradises” and new city designs.
However, experts warned the technology provided potential for bad actors to abuse it, with people sharing AI-altered images of a collapsed Eiffel Tower and a sinkhole swallowing the Great Pyramid of Giza.
A spokesperson for Google Earth said that they know that people “uniquely trust” the site to be a “reliable view of the world”.
They added: “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.”
The retracted feature allowed users to use text prompts to generate fake scenarios on top of a base layer of real satellite, aerial or 3D imagery.
The images could then be downloaded and shared on other platforms.
Dr Bissie Anderson, assistant professor in digital media and communications at the University of Birmingham, warned when Google Earth rolled out the feature that it could have “catastrophic implications for information integrity and the health of the digital information ecosystem”.
Anderson, who previously worked as a lecturer at Stirling and Robert Gordon universities, used the work of investigative open-source data journalist Henk Van Ess to show that “anyone can now plant a fake nuclear plant in Iran or place refugees on the Mexican border at the click of a button”.
Van Ess posted a warning about the new feature, saying: “Tonight I typed just one sentence into Google Earth and put refugees near the Mexican border. Then I planted a nuclear plant in Iran. What on earth is Google doing?”
The National also tested the technology when it was rolled out on Thursday using a basic free version of Google Earth which has fewer capabilities than paid options.
We successfully created several images relating to tanks taking over the streets of Glasgow or buildings close to The National's office being attacked by an airstrike, with the AI being asked to change the colour scheme of the satellite imagery to depict a different landscape (shown below).
Anderson explained that with the speed media can be shared on social media, this near-instant ability to create fake imagery could “unleash a wave of disinformation that would be near-impossible to debunk, and the SynthID watermark that synthetically generated Google content carries is small comfort”.
She added: “How many non-specialists outwith the OSINT, journalism or media research community, for instance, would be aware that such a tag means fake content?
“We simply cannot expect every single member of the public to possess the necessary level of media literacy or to be vigilant enough to be able to spot that.”
