Here’s the problem with putting an AI image generator in Google Earth

在 Google Earth 中加入 AI 图像生成器的问题所在

Even with Google’s watermarks and restrictions, a tool that created believable flyover images was ripe for misuse, as Henk van Ess demonstrated.

即便有谷歌的水印和限制，正如 Henk van Ess 所展示的那样，这种能够生成逼真航拍图像的工具极易被滥用。

A text prompt was all it took to generate reality-warping images using Google Earth’s satellite, aerial, and 3D imagery with a now-rolled back AI feature, like these images generated by Digital Digging’s Henk van Ess that show “refugees near the Mexican border” and a bomb crater near a hospital in Gaza.

只需一段文字提示，就能利用 Google Earth 的卫星、航拍和 3D 影像生成扭曲现实的图像。这项 AI 功能目前已被撤回，它曾生成过一些图像，例如 Digital Digging 的 Henk van Ess 所展示的“墨西哥边境附近的难民”以及加沙一家医院附近的弹坑。

Google initially responded to Digital Digging’s AI-altered images, saying, “We take misinformation seriously – every image created with Nano Banana in Google Earth includes the SynthID digital watermark, so if someone is unsure about an image, they can ask the Gemini app or use Lens in Search to see if the image was AI-generated.”

谷歌最初针对 Digital Digging 的 AI 修改图像回应称：“我们非常重视虚假信息问题——每一张通过 Google Earth 中 Nano Banana 功能生成的图像都包含 SynthID 数字水印。因此，如果有人对某张图像存疑，可以询问 Gemini 应用或使用搜索中的 Lens 功能，以查看该图像是否由 AI 生成。”

In addition to using Google’s “@verifyai” tag in Gemini to check suspicious images, van Ess suggests comparing them to other satellite platforms, like Sentinel-2 or Landsat, and even verifying orbital data like when it says the image was taken and by which satellite.

除了使用谷歌 Gemini 中的“@verifyai”标签来核查可疑图像外，van Ess 还建议将其与其他卫星平台（如 Sentinel-2 或 Landsat）进行对比，甚至核实轨道数据，例如图像拍摄的时间以及是由哪颗卫星拍摄的。

The feature launched globally on the web version of Google Earth earlier this week, and Google’s blog post pitched it as a tool for visualizing historical sites and real estate projects. Some images generated with it were easy to spot as fakes, like one showing the Sphinx crossed with the Statue of Liberty, but I can see how the Nano Banana 2 feature could have created images believable enough to spread misinformation.

该功能于本周早些时候在全球网页版 Google Earth 上线，谷歌的博客文章将其宣传为可视化历史遗迹和房地产项目的工具。其中生成的一些图像很容易被识别为伪造，例如一张狮身人面像与自由女神像合成的图片，但我能看出 Nano Banana 2 功能确实可能生成足以传播虚假信息的逼真图像。

And while Google’s watermark may be difficult to remove, it’s not impossible to miss or bypass – Digital Digging was able to fool Hive’s AI detector with an AI-altered video from Google Earth.

虽然谷歌的水印可能难以移除，但并非不可能被忽略或绕过——Digital Digging 就曾利用一段经 AI 修改的 Google Earth 视频成功欺骗了 Hive 的 AI 检测器。

Update, July 31st: After publication, Google announced it’s rolling back the feature.

更新，7 月 31 日：本文发布后，谷歌宣布将撤回该功能。