Google Pulls Earth AI Tool After Misinformation Risks Surface

Google has withdrawn a generative AI feature from the web version of Google Earth after researchers demonstrated its potential to create convincing misinformation. The tool, available for only 24 to 48 hours between July 30 and July 31, 2026, allowed users to overlay AI-generated imagery onto genuine satellite maps and coordinates.

Powered by the Nano Banana 2 AI model, the feature enabled users to weld invented visuals to real-world aerial and 3D mapping data. The rapid removal followed warnings from journalists and researchers who showed the tool could be used to fabricate fake warzones and disaster sites. Google stated the rollback was necessary to implement stronger guardrails for the technology.

The Credibility Gap

Google Earth is widely viewed as a primary, trusted source for global geospatial intelligence. By integrating generative AI directly into this platform, Google created a unique risk: fake images inherited the inherent credibility of the map's coordinates. Unlike standalone AI images, these fabrications were grounded in real-world context, making them significantly more deceptive to the average viewer.

Henk van Ess, an OSINT researcher, noted that the danger lies in the platform's authority, stating, "The forgery does not have to look convincing on its own. It inherits the credibility of the map it was born on."

Risks to Geospatial Intelligence

This incident underscores the vulnerability of open-source intelligence (OSINT) to AI-driven misinformation. The ability to streamline the creation of fake satellite imagery could allow bad actors to fabricate evidence of nuclear facilities, war crimes, or natural disasters. Such tools are particularly dangerous in fast-moving conflict zones where verified information is scarce and decisions are made based on available imagery.

Critics were swift in their assessment of the launch. The X account OSINTtechnical described the move as "unbelievably irresponsible" and claimed it caused "immediate, catastrophic damage to the credibility of satellite imagery."

Failed Safeguards

To prevent misuse, Google utilized SynthID digital watermarking to identify AI-generated content. However, the safeguard proved insufficient. Tests conducted by BBC Verify and other researchers revealed that these watermarks could be bypassed, and attempts to verify the images using Gemini often resulted in inconclusive answers.

Google has not yet provided a timeline for when, or if, a revised version of the tool will return. For now, the industry is watching whether other mapping services will implement similar generative features and what standards of verification will be required to prevent the erosion of trust in satellite data.