19 hours ago
- Google Earth added AI image generation (Nano Banana) that lets users create photorealistic images at any real-world coordinates, blurring the line between genuine and fabricated imagery.
- The generated images inherit credibility from Google's authentic satellite imagery, making them harder to detect as fakes, especially outside Google's ecosystem where watermarks and detection tools are ineffective.
- The feature poses a serious risk for misinformation, as demonstrated by examples like planting a nuclear plant in Iran or creating fake war scenes that have already spread online.
- Google's safeguards—SynthID watermark and detection via Gemini—are insufficient because fakes are often shared as screen recordings or re-encodes, and detectors fail on such media.
- Historical trust in Google Earth as a reference for verifying images (e.g., MH17 investigation) is now undermined because the same platform can produce convincing falsehoods.
- Effective verification requires cross-referencing with independent satellite sources (Sentinel-2, Landsat) and checking orbital elements, not relying on Google's own tools.
- Governments once censored Google Earth to hide truths; now the danger is the ease of creating falsehoods that appear authentic, with no equivalent procedure to stop it.