Google pulled the plug on its Earth AI feature within 24 hours of launch after users flagged serious misinformation risks. The tool let anyone generate fake AI imagery and overlay it directly onto Google Earth maps, creating convincing but false satellite views of real locations.
The backlash came fast. Critics pointed out the obvious problem: bad actors could use the feature to spread false information about neighborhoods, cities, or specific addresses. Someone could fabricate images showing flooding that never happened, construction that doesn't exist, or environmental damage that's purely fictional. When overlaid on actual maps tied to real GPS coordinates, these deepfakes gain credibility they shouldn't have.
Google hadn't announced a formal launch date for the feature, but it quietly appeared in Google Earth's web version. The company made its decision to remove it without releasing an official statement explaining why, though the timing and context made the reasoning clear.
This represents a rare moment of restraint from Google on an AI product. The company has pushed aggressively into generative AI across Search, Gmail, and other products despite ongoing questions about accuracy and hallucinations. But combining AI image generation with location data proved too risky to ship, even for a company that typically launches and iterates.
The Earth AI debacle highlights a core tension in the current AI boom. Generative tools excel at creating convincing content fast. That same capability becomes dangerous when tied to real-world locations and presented as factual information. Google likely saw internal testing that revealed how easily users could weaponize the tool for misinformation campaigns.
The company hasn't said whether it plans to relaunch a safer version. For now, Earth AI joins a small list of AI features that proved too problematic to release into the wild.