Skip to content

Inspect. Detect. Verify.

The AI detector that shows its evidence

Inspect text, images, documents, websites, source code and GitHub repositories for signs of AI generation - with provenance checks, metadata evidence and honest confidence levels.

  • Free, unlimited text checks - no account needed
  • Free website & GitHub repository scans
  • Private document, image & code scans on Pro

0 / 200,000 characters

Results are estimates with evidence - never proof of authorship.

Built for decisions you can defend

Model likelihood, done honestly

Scores come only from the detection model you configure. Without one, we say “unavailable” instead of inventing a number.

C2PA provenance

Read Content Credentials embedded by cameras and creative tools, including AI-generation declarations.

Metadata evidence

Generator signatures from Stable Diffusion, ComfyUI, Midjourney, DALL·E, Firefly and AI site builders.

Section-level findings

Paragraph, page, section and line references so reviewers can verify every claim themselves.

Secure by default

SSRF-protected fetching, content-verified uploads, archive-bomb protection and code that is never executed.

REST API

On Pro and Team: a versioned API with keys, rate limits, usage tracking and an OpenAPI specification.

How it works

  1. Submit content

    Paste text, enter a URL, connect a public repository or drop a file. Uploads are validated by their content, not just their extension.

  2. Asynchronous inspection

    Your inspection is queued and processed in the background with progress updates - large repositories and documents never block your browser.

  3. Separate evidence streams

    Model likelihood, C2PA provenance, metadata and heuristic signals are produced by independent engines and kept apart.

  4. Review and share

    Read the findings with highlighted passages and line references, export PDF or JSON - and on Pro and Team, compare inspections and share expiring links.

A report you can read in a minute

The headline likelihood is always accompanied by its reliability, the engines that produced it and the evidence behind it. Provenance and metadata are shown separately, and every report ends with its limitations.

How to read a report

Illustrative example - values are for layout only

AI-generation likelihood

62/100

Mixed signalsModerate reliability
  • highParagraph 4 scored 88 by the configured model
  • mediumContent Credentials present (signature not verified)
  • infoUniform sentence length - heuristic, not scored
  • infoLimitations: short sections reduce reliability
Every report separates model likelihood, provenance, metadata and heuristic evidence, and lists its limitations.

Plans for every workflow

Start free. Go Pro for private scans, Compare and the REST API. Compare plans

Free

For checking text, public websites and GitHub repositories.

$0forever

Public website & repo scans

50 inspections / month

  • Unlimited text checks
  • Website & GitHub repository scans
  • No account needed to start
  • Metadata & provenance checks
  • PDF & JSON reports
  • Results listed publicly in Explorer
Start free

Team

For organisations building inspection into their workflow.

Customvolume pricing

Private scans

20,000 inspections / month

  • Everything in Pro
  • REST API with 25 keys
  • Custom detection providers
  • Higher limits
  • Dedicated support
Talk to sales

Start inspecting in seconds

Check text, public websites and GitHub repositories for free - no account needed. Go Pro for private document, image and code inspections, Compare and the API.