# AI Video Detection API

> Detect AI-generated and deepfake videos with the Copyleaks API. Independent audio and visual analysis, time-based detection, and C2PA provenance.

Protect your organization from AI-generated video fraud. The [AI Detector](https://copyleaks.com/ai-detector) video API analyzes both the audio and visual tracks independently, providing granular time-based detection across the full video timeline - from deepfake evidence to synthetic media campaigns.

<Card title="Try it in the Playground" icon="square-terminal" href="https://api.copyleaks.com/dashboard/playground/video-detection" arrow="true">
  Submit a video and see the audio + visual breakdown live, no integration required. Opens in a new tab.
</Card>

![AI-generated deepfake example showing a synthetic face swap](/images/ai-generated-deepfake-feature-v2-1024x509.webp)

## The risks of synthetic videos

AI video generation went from creative tool to security threat in a year. Anyone can produce photorealistic videos of people, places, and events that never happened. For enterprises that creates real exposure:

- **Legal evidence** - courts need verification that video evidence isn't artificially manufactured or edited
- **Insurance fraud** - synthetic accident scenes and AI-generated footage threaten claim integrity
- **Financial crime** - deepfake executive videos enable sophisticated social engineering
- **Disinformation** - AI-generated news footage undermines public trust
- **Regulatory compliance** - industries face growing requirements to verify media authenticity
- **Reputation protection** - synthetic video targeting executives or brands can cause irreversible damage

## Capabilities

<CardGroup cols={2}>
  <Card title="Audio & visual track analysis" icon="video">
    Independently analyzes both the audio and visual tracks, returning separate AI ratios and time-based detection data for each.
  </Card>
  <Card title="Time-based detection" icon="clock">
    Returns start positions and durations (in milliseconds) for each AI-detected segment, enabling frame-accurate analysis across the full timeline.
  </Card>
  <Card title="Overall AI ratio" icon="gauge">
    Provides a single `overallAIRatio` score representing the proportion of the video (union of audio and visual AI detections) that is AI-generated.
  </Card>
  <Card title="Metadata extraction" icon="file-export">
    Extracts embedded provenance metadata (C2PA standard) to identify the generating tool, issuing organization, and creation timestamp when available.
  </Card>
  <Card title="Broad format support" icon="file-video">
    `.mp4`, `.avi`, `.mov`, `.mkv`, `.webm`, `.flv`, `.wmv`, `.mpg`, `.m4v`, `.3gp`, `.mxf` - up to 512 MB and 1 hour in duration.
  </Card>
  <Card title="Async webhook delivery" icon="plug">
    Designed for large files: submit a video URL and receive results via webhook when processing completes, no polling required.
  </Card>
</CardGroup>

## Use cases

<Tabs>
  <Tab title="Legal & Forensics">
    Verify authenticity of video evidence in legal proceedings and investigations. Ensure that footage submitted to courts or regulatory bodies hasn't been artificially generated or manipulated.
  </Tab>
  <Tab title="Insurance Fraud Prevention">
    Detect synthetic accident footage, AI-generated property damage videos, and fake evidence in insurance claims.
  </Tab>
  <Tab title="Media & Journalism">
    Verify authenticity of news footage, user-submitted videos, and source material. Protect editorial integrity by ensuring video content hasn't been artificially generated.
  </Tab>
  <Tab title="Financial Services Security">
    Detect deepfake executive videos used for social engineering, synthetic identity verification footage, and AI-generated proof-of-life recordings.
  </Tab>
  <Tab title="Enterprise Governance">
    Monitor for synthetic media targeting your brand, executives, or communications channels. Safeguard reputation and prevent disinformation campaigns.
  </Tab>
  <Tab title="Identity Verification">
    Deepfakes are a direct threat to identity verification workflows. AI-generated videos can convincingly impersonate real individuals - enabling attackers to bypass video KYC, spoof liveness detection, and fabricate proof-of-identity recordings.

    Use the AI Video Detection API to verify that submitted identity videos are genuine before approving onboarding, access requests, or high-value transactions. Detect synthetic faces, AI-cloned voices, and manipulated footage across the full video timeline.
  </Tab>
</Tabs>

## Pair with other detection APIs

<CardGroup cols={3}>
  <Card title="AI Image Detection" icon="image" href="/concepts/products/ai-image-detection-api">
    Detect AI-generated images to complement video verification for complete media authenticity.
  </Card>
  <Card title="AI Text Detection" icon="robot" href="/concepts/products/ai-text-detection-api">
    Detect AI-generated text content for complete document and media authenticity.
  </Card>
  <Card title="Plagiarism Detection" icon="magnifying-glass" href="/concepts/products/plagiarism-checker-api">
    Verify that content hasn't been copied from other sources alongside AI generation checks.
  </Card>
</CardGroup>

## Frequently asked questions

### Can it detect deepfakes?

Yes. The API detects AI-generated and deepfake video, analyzing the full timeline to flag synthetic faces, AI-cloned voices, and manipulated footage - useful for fraud prevention, forensics, and identity verification.

### Does it analyze audio and video separately?

Yes. It independently analyzes the audio and visual tracks, returning separate AI ratios and time-based detection data for each, plus a single `overallAIRatio` for the whole video.

### Which video formats and sizes are supported?

`.mp4`, `.avi`, `.mov`, `.mkv`, `.webm`, `.flv`, `.wmv`, `.mpg`, `.m4v`, `.3gp`, and `.mxf`, up to 512 MB and 1 hour in duration.

### Is video detection synchronous?

No. It is designed for large files: you submit a video URL and receive results via webhook when processing completes, with no polling required.

### Can it identify which tool generated a video?

When embedded provenance metadata is present, it extracts C2PA data to identify the generating tool, issuing organization, and creation timestamp.

## Next steps

<CardGroup cols={2}>
  <Card title="Start Building" icon="rocket" href="/guides/ai-detector/ai-video-detection">
    Get started with the API and integrate AI video detection into your workflows.
  </Card>
  <Card title="Endpoint Reference" icon="code" href="/reference/actions/ai-video-detector/submit">
    Technical specifications, request parameters, and webhook payload reference.
  </Card>
</CardGroup>

<Card title="Detect AI-Generated Videos at Scale" icon="calendar-check" href="https://copyleaks.com/book-a-demo" cta="Request demo" arrow="true">
  Schedule a personalized demonstration with our enterprise security team. We'll analyze your specific risk scenarios and show you how detection prevents fraud in your industry.
</Card>