谷歌在Google Earth中推出了一款名为Nano Banana的AI图像生成工具，允许用户通过文本提示在卫星影像、航拍照片和3D模型的基础上生成图像[1]。然而，研究人员已利用该工具生成了包括"墨西哥边境难民"和"加沙医院附近弹坑"在内的具有误导性的图像，引起了人们对虚假信息传播的担忧[1]。
为应对这些风险，谷歌表示所有由Nano Banana生成的图像都包含SynthID数字水印[1]。用户可通过Gemini应用或Search Lens验证图像是否由AI生成[1]。
Google has integrated an artificial intelligence image generation tool called Nano Banana into Google Earth, enabling users to create images based on satellite imagery, aerial footage, and 3D data through text prompts [1]. The tool generates what researchers describe as "reality-warping" images derived from the underlying geographic data [1].
Researchers have already demonstrated the potential for misuse by generating misleading images through the platform [1]. Henk van Ess from Digital Digging used Nano Banana to create deceptive visuals, including scenes depicting refugees near the Mexican border and bomb craters near a Gaza hospital [1]. These examples highlight concerns about the tool's capacity to produce false or distorted representations of real-world locations [1].
To address misinformation risks, Google has implemented safeguards for the generated content [1]. According to the company, all images produced by Nano Banana contain SynthID digital watermarks that enable verification [1]. Users can validate whether an image was AI-generated by checking it through Google's Gemini application or Search Lens [1].
