Made Mark — An open specification for content authorship disclosure
An open specification for content authorship disclosure.
What is Made Mark?
Made Mark is an open specification for content authorship disclosure. It answers the question no existing specification addresses: how was this content made? A CC license tells you what you can do with content. CC Signals tells you whether it can train AI. Made Mark tells you who — or what — made it.
It is built on the same three-layer architecture as Creative Commons licenses — a visual mark, a plain-language label, and a machine-readable expression. Made Mark is designed to travel inside C2PA manifests as a human-readable semantic layer, giving authorship disclosure both cryptographic provenance and plain-language meaning. Standalone use without C2PA is also supported.
The Made Mark specification is public domain. Use it, implement it, build on it.
The three marks
Conceived and produced by people. No AI tools involved in the creative work itself.
mademark.org/marks/human-madeHuman creative direction, AI production. A person shaped the intent — AI produced the artifact.
mademark.org/marks/human-designed-ai-madeConceived and produced primarily by AI, with no human creative direction for the specific output.
mademark.org/marks/ai-madeThe hallmark lineage
For seven hundred years a hallmark has stamped who made an object and to what standard — from the silversmith's punch to the Roycroft mark in East Aurora. Made Mark brings that idea to digital content.
Three base marks, kept few. Modifiers for model, provider, and source license ride alongside — optional, not required.
The label
Made Mark produces a structured plain-language label with a full form and a short form.
The mark is the only required field. All others are optional and shed as space demands.
Short form
The short form keeps the system identifier and the mark — the irreducible core. It drops model, provider, and source license when space is tight.
Machine layer
Made Mark is built in three layers, like a CC license. Each layer serves a different audience. Together they make authorship disclosure work for both people and software.
Visual mark
A badge anyone recognizes at a glance.
Plain-language label
MM · Human Designed, AI Made · Claude Sonnet 4.6 · Anthropic
Machine-readable expression
Stable URIs plus structured metadata that software can parse automatically.
Mirrors CC's own deed / legal code / metadata structure.
Schemas and integration
Made Mark expresses as JSON-LD for web pages, XMP for image, video, and audio files, and as a custom assertion inside C2PA manifests. Copy-paste schemas, stable URIs, badge assets, and platform guidance live on the Implement page.
Machine-readable from day one. The metadata rides inside C2PA's manifest, so Made Mark travels with provenance instead of competing for it.
Use and licensing
Made Mark is built on a split model — the same approach Creative Commons uses for its own work. The openness is in the specification. The protection is in the signal.
CC0, public domain. The vocabulary, label syntax, URI structure, and documentation are dedicated to the public domain. No rights reserved. Implement Made Mark without asking.
CC BY-ND 4.0. Free to use unmodified to identify Made Mark authorship on content. Do not modify, recolor, or repurpose outside this context. Credit: Made Mark, mademark.org.
"Made Mark" and the [m] wordmark are trademarks of Daniel Richard Skrok (USPTO application filed June 23, 2026, Class 42). They exist to protect the integrity of the signal. You do not need permission to use Made Mark on your content. You may not use the name or logo to represent a competing or derivative specification.
Made Mark is currently administered by Daniel Richard Skrok. A governance policy covering administration, misuse handling, and conditions for stewardship transfer to a standards body is published at mademark.org/governance.