Google pauses AI satellite images, after fears of deepfakes in the sky

Google recently removed a feature from its Earth platform that allowed users to create AI-generated satellite images. The tool enabled anyone to type a prompt and receive a synthetic image of locations around the world. The company stated that the rollback is temporary while they implement stronger protections after reports showed the feature was used to create misleading imagery.

Journalists and open-source investigators raised immediate concerns about the potential for abuse. Researchers found it was simple to generate fake images of sensitive sites, including nuclear plants and border zones. The ability to create credible satellite fakes with a few keystrokes presents significant challenges for those who rely on satellite data to verify breaking news and document world events.

Open-source experts emphasized that satellite imagery has been a vital, high-trust tool in journalism because it is difficult to manipulate. The introduction of accessible generation tools risks undermining that trust, as bad actors can now produce images that look authentic. Investigators warn that these capabilities will accelerate the spread of misinformation if not strictly controlled.

Google maintains that it takes misinformation seriously and noted that all generated images contain a digital watermark to identify them as AI creations. However, the open-source community remains skeptical, noting that the tool essentially compromised one of the primary methods used to check the truth of online claims. The company is now working to update its safeguards before considering any future release of similar capabilities.