跳到主要内容
Learn Motion
探索课程产品能力价格

下载 Learn Motion

用手机扫描二维码,下载 App。

App Store

扫码从 App Store 下载 Learn Motion打开商店页面

Google Play

扫码从 Google Play 下载 Learn Motion打开商店页面
登录
Learn Motion

Spotting AI-Generated Images and Videos: A Practical Checklist

1Why Detection Is a Judgment, Not a Test2Visual Clues in AI-Generated Images3Visual Clues in AI-Generated Video4Context and Content Clues5Metadata and Provenance Signals6A Practical Checklist and Common Traps
Metadata and Provenance Signals

Reading a File's Metadata

1 / 3
Think of a photo file as two layers stacked together. The top layer is the picture you see. Underneath is a block of hidden text called metadata, and for photos the common format is EXIF. It records the camera model, the exposure settings, the capture time, and sometimes GPS coordinates. When you open a file's properties, those fields are what you read. A specific camera model and an original timestamp are consistent with a real capture. But two cautions matter. Most social platforms strip metadata on upload, so a file saved from a post usually has none, and that absence tells you nothing about whether AI made it. And metadata is easy to edit, so a clean-looking EXIF block can be fabricated. Treat every field as a weak signal, not a verdict.
0:00 / 0:00

A digital photo or video file usually contains two parts: the visible image data and a hidden block of descriptive data called metadata. For photos, the most common format is EXIF (Exchangeable Image File Format), which records the camera make and model, the exposure settings, the date and time of capture, and often GPS coordinates. Video files carry equivalent information in container-level tags.

Reading metadata is a matter of opening the file's properties or using a metadata viewer. The fields that matter most for a quick judgment are:

  • Camera make and model: a specific camera model is consistent with a real capture; a blank or generic value is common in AI output.
  • Date and time: an original capture timestamp is useful for cross-checking against the claimed event date.
  • GPS coordinates: if present, they can be checked against the location claimed in the caption.
  • Software field: this often names the editing or generating application. A value such as "Adobe Photoshop" indicates editing; some generators write their own name here.

Two cautions apply. First, most social platforms strip metadata automatically on upload, so a file downloaded from a social post will usually have none — that absence is not evidence of AI generation. Second, metadata is trivially editable: free tools can rewrite any field, so a plausible-looking EXIF block can be fabricated.

上一页1 / 3下一页

Learn Motion

生成课程,认真学会。

运营方:Wuhan Daoyin Technology Co., Ltd.

联系我们: [email protected]
隐私政策用户服务协议

© 2026 Learn Motion