> ## Documentation Index
> Fetch the complete documentation index at: https://docs.dataspike.io/llms.txt
> Use this file to discover all available pages before exploring further.

# What We Detect

> Attack vectors covered by the Deepfake API across images, video, and audio — and the fraud scenarios they map to.

Coverage depends on the media type. Images and video are analyzed for **face manipulation**; audio is analyzed for **voice synthesis and cloning**.

***

## Attack vectors by media type

| Attack vector                  | Image | Video | Audio |
| ------------------------------ | :---: | :---: | :---: |
| Face swap                      |   ✅   |   ✅   |   —   |
| Manipulated faces              |   ✅   |   ✅   |   —   |
| Synthetic / AI-generated faces |   ✅   |   ✅   |   —   |
| Talking-head deepfakes         |   —   |   ✅   |   —   |
| Lip-sync manipulation          |   —   |   ✅   |   —   |
| Face reenactment               |   —   |   ✅   |   —   |
| Voice cloning                  |   —   |   —   |   ✅   |
| Synthetic speech               |   —   |   —   |   ✅   |

***

## Fraud scenarios

<CardGroup cols={2}>
  <Card title="KYC impersonation" icon="user-secret">
    Passing verification on behalf of someone else.
  </Card>

  <Card title="Synthetic account creation" icon="users">
    Mass sign-ups using AI-generated faces.
  </Card>

  <Card title="Hiring fraud" icon="briefcase">
    Substituting a candidate during a remote interview.
  </Card>

  <Card title="Proctoring bypass" icon="graduation-cap">
    Defeating online exam supervision.
  </Card>

  <Card title="Voice spoofing" icon="microphone">
    Beating voice-based authentication.
  </Card>

  <Card title="Fraud calls" icon="phone">
    Calls placed with a cloned voice.
  </Card>
</CardGroup>

<Note>
  Beyond neural detection, every check also runs a [metadata & forensic layer](/deepfake-detection/references/metadata-forensics) that looks for traces of editing and AI generation.
</Note>