On July 30, Google introduced a new AI-powered "Create Image" feature in Google Earth that allowed users to zoom into virtually any location, enter a text prompt, and generate photorealistic scenes layered over real satellite imagery, aerial photography, and 3D map data. Powered by Google's latest image generation model, the feature aimed to combine creative AI with one of the world's most trusted mapping platforms.
Google confirmed it was pulling the feature almost as quickly as it had rolled it out, saying that while many users, particularly geospatial professionals, had put it to legitimate, creative use, others had shared generated images that appeared to cross the company's content policies. Rather than let the situation escalate, Google chose to hit pause.
"We've seen geospatial professionals using this feature for a range of useful purposes," the company said in a statement, "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."
What made the walk-back notable wasn't just the speed. It was what Google Earth represents. For nearly two decades, the platform has been a go-to reference for journalists and open-source investigators trying to verify what's really happening on the ground, from disaster zones to conflict areas. Its imagery has long served as a rare anchor of visual truth in an internet increasingly crowded with fabricated content.
That's precisely why critics reacted so quickly. Researchers and disinformation experts warned that letting users generate convincing fictional scenes over real satellite imagery of actual landmarks and monuments could blur the line between documentation and fabrication, a risk that felt especially uncomfortable coming from a platform people specifically turn to because it's considered hard to fake.
Google, for its part, noted that the AI-generated images never appeared in the main Google Earth experience and were clearly watermarked as artificially created. Still, screenshots of the output began circulating, and some of it apparently didn't sit well with the company's own guidelines.
Google isn't alone in learning this lesson the hard way. As tech giants race to bolt generative AI onto nearly every product they ship, plenty have found themselves yanking features back almost as fast as they launched them, usually after users find creative, unintended ways to break them. Just weeks earlier, another major platform discontinued a similar AI image tool tied to public social media profiles after a wave of privacy backlash.
The pattern is becoming familiar: ship fast and be ready to reverse course the moment things go sideways.
Google says the feature isn't dead, just paused while it builds "stronger guardrails" around what can and can't be generated. Exactly what that will look like, or when the tool might return, hasn't been detailed yet.
Comments