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

Google had a rough 24 hours with its mapping app. On Thursday, the company slipped Nano Banana 2, its AI image generator, into Google Earth, letting anyone superimpose fabricated scenes onto real satellite maps. The pitch was playful: get creative with geography, dream up alternate versions of your neighborhood or some far-off coastline. By Friday afternoon, the feature was gone.

The backlash arrived fast, and it came from people who actually rely on Google Earth for their jobs. Journalists and researchers have long treated the app as a trustworthy visual record, a place where what you see is, at minimum, tied to something real. A BBC reporter summed up the mood with a sarcastic post noting that surely nothing bad could come from bolting an AI image generator onto one of the internet's most cited sources of geographic evidence. That sarcasm turned out to be the whole story.

Google's own explanation for the reversal is telling. The company said geospatial professionals found genuine uses for the tool, but also acknowledged seeing screenshots of AI-generated imagery that violated its policies circulating online. In other words, the fabricated maps were already leaking out of the sandbox and into feeds where context gets stripped away fast. Google says it's rolling the feature back while it builds stronger guardrails, though it hasn't said what those guardrails will look like or when the tool might return.

What makes this episode more than a one-day news cycle is the timing. Anyone with access to a modern AI image generator, and there are now several competing ones on the market, can already fake convincing satellite-style imagery without needing Photoshop chops. Google didn't invent that capability. It just packaged it inside an app people have trusted for two decades, which made the risk far more concrete and far easier to spot.