Enterprise Grade • Open Source

Remove AI Provenance
Marks from Any File

Remove invisible Unicode watermarks, C2PA / EXIF / XMP metadata, and statistical text watermarks from your documents, images, and containers — covering all major AI vendors.

Claude Gemini / SynthID OpenAI Kirchenbauer-style
14+ File Formats
3 Watermark Layers
4 AI Vendors
How It Works

Three Layers of Protection

A comprehensive pipeline eliminates every type of AI provenance mark from your content.

Layer A
A

Invisible Unicode

Deterministic removal of zero-width chars, exotic spaces, bidirectional control characters, and Unicode tag chars.

Files
F

File Metadata

Remove C2PA manifests, EXIF tags, XMP blocks, and document properties from 14+ file formats including images, PDFs, and Office docs.

Interactive Tool

Inspect & Clean Your Files

Upload a file or paste text to securely remove AI provenance marks.

Drop a file here or browse

PNG • JPEG • WebP • PDF • DOCX • ODT • EPUB • SVG • HTML • MD • TXT • and more

Developer API

REST API Endpoints

Integrate our powerful cleaning engines directly into your workflows with our scalable API.

GET
/health

Liveness probe — returns version string

{"ok": true, "version": "0.5.0"}
GET
/capabilities

Available tools and optional backends

{"tools": {"c2patool": true},
 "layer_b": {"available": false}}
POST
/inspect

Detect AI provenance marks; returns findings report

POST {"file": "<base64>", "name": "notes.md"}
→ {"ok": true, "suspicious": true, "report": {...}}
POST
/clean

Remove all marks; returns cleaned bytes + report

POST {"file": "<base64>", "name": "doc.txt",
      "options": {"layer_b": true}}
→ {"ok": true, "cleaned": "<base64>"}