Google Suspends AI Tool in Google Earth After Users Create Fake Satellite Images
Translated & summarized from Mako by baba
Google suspended a new AI image creation tool in Google Earth less than two days after launch due to misuse concerns. Users generated fake satellite images depicting scenarios like an Iranian nuclear plant and a Gaza hospital bombing. The company is working on stronger safeguards as experts warn the tool could spread convincing misinformation.
The story in 6 lines · by baba
- Google suspended its new AI image creation tool in Google Earth within 48 hours of launch due to misuse.
- Users created fake satellite images including an Iranian nuclear facility and Gaza hospital bombing.
- None of the AI-generated images appeared on Google Earth’s main platform, Google said.
- Researcher Henk van Ess warned the tool could enable misleading images that inherit map credibility.
- BBC fact-checkers produced convincing fake images, exposing detection tool weaknesses.
- Google is developing stronger safeguards before reintroducing the feature.
Less than 48 hours after launching a new AI-powered image creation tool in Google Earth, Google announced it has suspended the feature due to concerns over misuse and the spread of misinformation. The tool allowed users to generate fictional scenarios overlaid on real satellite imagery, aerial photos, and 3D models using text prompts. Among the fabricated images created were a nuclear facility in Iran and a hospital bombing in Gaza.
Google stated in a post on X that none of the AI-generated images appeared on the main Google Earth platform. While the company acknowledged the tool’s potential for beneficial professional use, it also noted that users shared screenshots of images that appeared to violate company policies. Dutch intelligence researcher Henk van Ess warned immediately after the tool’s release that malicious actors could exploit it to create convincing fake images, complicating the work of journalists, researchers, and fact-checkers.
Van Ess highlighted that the AI-generated images inherit the credibility of the underlying Google Earth maps, making it harder to distinguish real from fabricated content. He cited examples such as nonexistent Iranian facilities, refugee camps on the US-Mexico border, and bombed hospitals in Gaza. Tests revealed that some of Google’s built-in restrictions could be bypassed by rephrasing prompts, allowing the creation of images that should have been blocked.
BBC’s fact-checking team also demonstrated the tool’s vulnerabilities by producing fake images of a collapsed Eiffel Tower, a sinkhole beneath the Great Pyramid of Giza, and Russian tanks in Kyiv. Google said AI-generated content includes invisible watermarks to indicate manipulation and recommended users verify suspicious images using tools like Gemini or Google Lens. However, BBC Verify found these detection methods were not foolproof, with some AI tools misidentifying fake images as authentic.
Google emphasized it is suspending the feature "while working on stronger safeguards". The company reiterated that the images did not appear on Google Earth’s main platform and acknowledged the challenge of integrating advanced AI tools into products relied upon for trustworthy information. This move reflects the broader difficulty of balancing AI innovation with preventing misinformation.
Mentioned
