Google nixes its Earth AI feature one day after launch, amid criticism it would spread misinformation

In Brief

Google rolled retired a caller feature connected Thursday that allowed users to deploy Nano Banana 2, its AI representation generator, to make fabricated images wrong its outer imagery mapping app Google Earth. The full constituent of this feature, Google said, was to get originative with geography.

But critics quickly pointed out that the instrumentality could beryllium utilized to make and dispersed misinformation. The tool, which was prompt-based and allowed beauteous overmuch immoderate representation to beryllium superimposed implicit existent maps, seemed similar a look for a deluge of geospatial slop.

“There’s nary mode that this caller AI representation procreation diagnostic connected Google Earth, 1 of the astir reliable sources of ocular grounds for journalists and researchers, could perchance beryllium abused to dispersed misinformation online,” a BBC writer posted sarcastically Thursday.

Now, lone a time aft the feature’s release, Google has scrapped it.

“We’ve seen geospatial professionals utilizing this diagnostic for a scope of utile purposes, nevertheless we’ve besides seen radical sharing screenshots of generated imagery that look to interruption our policies,” the institution said successful a statement.

“We’re rolling backmost this diagnostic successful Google Earth portion we enactment connected implementing stronger guardrails,” Google added.

The disapproval of Google’s instrumentality is understandable, though it’s worthy pointing retired that, successful the property of AI, astir imagery that exists connected the web tin beryllium easy manipulated. You don’t person to beryllium a Photoshop whiz anymore to make semi-credible disinformation, you conscionable request entree to an AI representation generator, which Google sells, arsenic bash galore different AI firms.

Subscribe for the industry’s biggest tech news