Google said on Friday it would be pausing a new AI feature it introduced to Google Earth which made it possible for users to manipulate satellite mapping images.
After initially seeming to dismiss concerns, the company issued a statement on its media relations account on X announcing the pause.
"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," the company said.
"So we’re rolling back this feature in Google Earth while we work on implementing stronger guardrails.“
Hours earlier, as criticism was mounting, the Alphabet-owned company stood by the feature.
An earlier statement said that the company takes misinformation “seriously”, highlighting that images created with Nano Banana in Google Earth included the SynthID digital watermark – a tool that makes it easier for users to identify content created with or enhanced by AI.
“If someone is unsure about an image, they can ask the Gemini app or use Lens in Search to see if the image was AI-generated,” the company added, also pointing out that Google “continually updates” guardrails that prevent the creation of certain images that might be deemed inappropriate.
Nano Banana is Google's image-editing layer for its Gemini AI tool, and it has become one of the more popular image-editing and generation apps.
Concerns began to mount after users discovered how easy it was to manipulate satellite images that could work in tandem with politically charged world events.
“Tonight I typed just one sentence into Google Earth and put refugees near the Mexican border,” Henk Van Ess, a technology and AI research expert, wrote on X. “Then I planted a nuclear plant in Iran. What on earth is Google doing?”
The post, linked to a longer article explaining his concerns about Google's recently added feature, has been reposted more than 1,200 times and received more than 2.4 million views on X alone.
With only one prompt, The National was able to make it appear as though the US Capitol building in Washington was surrounded by data centres.
Google's response to Mr Van Ess, however, has been met with criticism, with many pointing out that the company seems to be advising users to employ its own tool, SynthID, to identify manipulated images on its mapping tool, Google Earth. Others said the tool had limited scope.
“Google's SynthID only watermarks AI-generated portions of Nano Banana images in Google Earth (less than 25 per cent in tested examples), not the entire image, limiting full verification as AI-generated,” a context explanation appearing below Google's response to the criticism read.
SynthID, which Google introduced in 2023, has continued to be updated and integrated into the company's various offerings. In May, Google chief executive Sundar Pichai said OpenAI, Kakao and ElevenLabs had adopted SynthID's AI watermark standard to make it even more effective in flagging content.
Within the Google Earth web tool, the company has labels noting that the feature is experimental.
“Generative AI can make mistakes, so double check it,” a line of text reads near the area where users can enter prompts to manipulate the map image.
On Thursday, the technology giant highlighted the latest version of Google Earth in a blog post.
“Transform any place with Nano Banana in Google Earth,” the headline read, accompanied by an article showing examples of how the tool could be used to “bring history to life”, “create professional real estate plans” and “give your favourite place a makeover”.
Google Earth's AI prompt feature, however – much like OpenAI's Sora and similar content generation tools used by xAI's Grok – is likely to generate controversy, given recent polls showing increased concern about AI's impact on society.
Meanwhile, backlash against the feature has continued to gain momentum.
On Friday afternoon on the company's media release account on X, it reiterated
