Google Earth Briefly Offered an AI Image Generator That Created Fake Satellite Imagery — Then Rolled It Back
Summarized by AI from reporting by The Verge AI, published under our editorial policy.
Google Earth briefly offered an AI feature that could generate realistic but misleading satellite images from text prompts. The tool was quickly rolled back after researcher Henk van Ess demonstrated it could fabricate scenes like refugees at the Mexican border and bomb craters near Gaza hospitals, raising serious misinformation concerns.
Key takeaways
- Google Earth briefly offered an AI feature that generated satellite images from text prompts.
- Researcher Henk van Ess demonstrated the tool could create fake images of refugees near the Mexican border and a bomb crater near a Gaza hospital.
- Google rolled back the feature after recognizing its potential for spreading misinformation.
Google Earth recently introduced and then quickly rolled back an AI feature that allowed users to generate satellite images from text prompts. This tool, which used Google Earth's satellite, aerial, and 3D imagery, could create realistic but misleading images, such as showing "refugees near the Mexican border" or a bomb crater near a hospital in Gaza. The feature was highlighted by Digital Digging's Henk van Ess, who demonstrated its capabilities.
How the AI Feature Worked
The AI feature in Google Earth used text prompts to generate satellite images that appeared authentic. Users could input descriptions like "refugees near the Mexican border" or "a bomb crater near a hospital in Gaza," and the AI would create corresponding images. These images were based on Google Earth's extensive database of satellite and aerial imagery, making them look convincing. However, the generated images were not real and could easily mislead viewers about actual events or locations.
The Risks of Misleading Imagery
The primary concern with this AI feature was its potential to spread misinformation. Satellite images are often used as evidence in news reports, legal proceedings, and humanitarian efforts. The ability to generate realistic but fake images could undermine trust in visual evidence. For example, an image of a bomb crater near a hospital in Gaza could be used to falsely accuse a party of war crimes, or an image of refugees near the Mexican border could fuel political debates with fabricated visuals.
Why Google Rolled Back the Feature
Google quickly rolled back the AI feature after it became clear that the tool could be misused. The company likely recognized the ethical and legal implications of allowing users to create misleading imagery. While AI tools can offer innovative ways to explore and interact with data, they also come with significant risks, particularly when dealing with sensitive topics like geopolitical conflicts and humanitarian crises.
What You Can Do Today
If you use Google Earth, be cautious about the information you consume and share. Verify the authenticity of images, especially those that seem to depict sensitive or controversial events. You can also report any misleading or fake images you encounter to Google Earth's support team. Additionally, consider using other tools like Google's reverse image search to verify the origin of any questionable images you come across.
Frequently asked
- Is the AI feature in Google Earth still available?
- No, Google has rolled back the AI feature that generated satellite images from text prompts.
- How can I verify the authenticity of satellite images?
- You can use tools like Google's reverse image search to check the origin of any questionable images.
- Who discovered that Google Earth's AI could generate misleading satellite images?
- The feature was highlighted by Henk van Ess of Digital Digging, who demonstrated its capabilities.