Google pauses AI satellite images, after fears of…

In July 2026, Google introduced a new feature for its Google Earth platform that allowed users to generate AI-based satellite imagery through simple text prompts. The tool was quickly pulled back after researchers warned that it could facilitate the rapid creation of realistic deepfake satellite imagery, potentially undermining the use of such data for verifying news and global events.

What happened

The feature allowed users to type specific scenarios into the Google Earth web interface to generate corresponding satellite visuals. Open-source investigators, including Henk van Ess, successfully used the tool to create fabricated scenes such as a flooded U.S. Capitol complex and fires on Iran's Kharg Island. Experts noted that satellite imagery has historically been considered a reliable method for verifying ground-level events, but the accessibility of these AI tools threatened to compromise that trust.

Prior to this incident, fake satellite imagery had already appeared in the public sphere. During the conflict between the U.S.-Israel and Iran, manipulated images circulated in Iranian media showing damage to a U.S. military base in Bahrain. These fakes reportedly utilized existing Google Earth imagery combined with AI to simulate structural damage. While the base was indeed damaged, the specific details shown in the fake images were inaccurate. Researchers like Jake Godin of Bellingcat warned that integrating such generation capabilities directly into mapping software streamlines the production of misinformation, making it harder for the public to distinguish between authentic and synthetic evidence. Google stated that it is working on implementing stronger guardrails and noted that generated images include its SynthID watermark, though the open-source community expressed significant concern that the tool could be used to facilitate mass disinformation.