MarkMyAss

Claude Watermark Remover & AI Provenance Cleaner

Inspect Clean Verify

MarkMyAss shows exactly what it removed — and what it cannot verify.

Proof, not promises.

Inspect a File Explore the Lab
  • Free
  • Open source
  • No account
  • Independent verification
  • Download your cleaned file

No fake "100% undetectable" scores.

Illustration of a pirate crew aboard their ship: the captain locks a glowing spyglass scan-beam onto a startled ghost, a crewmate nets a fleeing ghost, another finds a ghost hiding in a treasure chest while laughing and pointing at a skull stamp mark on the net-thrower's trousers

Works beyond Claude

Remove supported AI watermarks, metadata and hidden file traces from your AI images, text, PDFs and documents — before you publish.

Whichever AI made it, MarkMyAss inspects, cleans and verifies the supported signals inside the file. Model-level statistical watermarks stay honestly marked UNKNOWN — never claimed removed.

Let's see what is hiding inside.

Supported: TXT · MD · JSON · CSV · JPG · JPEG · PNG · WEBP · PDF

Use MarkMyAss automatically

Install MarkMyAss in your AI workflow and let the crew inspect, clean and verify supported traces as you create.

Install the Skill →

What MarkMyAss actually removes

Signal-based, not provider-magic — support depends on the signal type, not just which AI made the file.

Reliably supported

  • Hidden Unicode and invisible text characters
  • EXIF / XMP / IPTC metadata (JPEG, PNG, WebP)
  • PDF document metadata (DocInfo + XMP)
  • PNG metadata
  • Supported file-level provenance signals
  • Independent metadata verification with ExifTool

Current limits

  • C2PA / Content Credentials support is partial
  • Claude, Gemini, and GPT statistical model-level watermarks are currently UNKNOWN — no public verifier exists to confirm removal
  • MarkMyAss never claims "100% AI-undetectable"

What "VERIFIED CLEAN" means

MarkMyAss shows VERIFIED CLEAN only when a supported signal was actually found before cleaning, its own re-inspection confirms it's gone, and every independent verifier that could run (ExifTool, and c2patool where applicable) agrees too. It is not a claim that the file contains no possible identifying signal whatsoever — only that the specific, supported metadata categories MarkMyAss targets are independently confirmed gone. See the AI Watermark Lab for the full, honestly-scored capability matrix.

Is MarkMyAss only for Claude?

No. MarkMyAss is not limited to Claude. It works across supported watermark and provenance signal types, including hidden Unicode and common file metadata. Its support depends on the signal type, not just the AI provider.

Best use cases

Where MarkMyAss is genuinely reliable — not a universal AI-detection eraser, but strong at what it actually does.

AI Watermark Lab

Chart the waters of AI provenance — the full capability matrix, honestly scored: verified, partial, or unknown.

Explore the Lab →

Benchmarks

Captain's Log — every supported test, every failure, generated fresh from the real pipeline. Nothing hidden.

Read the Log →

Run Models Locally

Sail your own stack — run open-weight AI models yourself and skip provider-side provenance at the source.

Sail your own stack →

From one tool to a whole crew

MarkMyAss cleans the hidden traces in your AI content. Moseisley gives you a personal crew of AI agents and assistants to help you work, research, plan and automate.

MarkMyAss cleans your AI files. Moseisley gives you the whole crew.

Build your AI crew — free →