cred — A memorable standard for Human-AI attribution

Table of contents

• 1. Purpose
• 2. Format
• 3. Contribution codes
• 4. Common examples
• 5. Attribution layers
• 6. Usage guidelines
• 7. Responsibility
• 8. Related initiatives
• 9. License

1. Purpose

The cred standard provides a simple, universal way to declare human and AI contributions in creating a work. Inspired by Creative Commons for copyright, it scales from quick shorthand to detailed attribution. The system is voluntary and trust-based: there is no oversight or compliance mechanism. The goal is to give authors a standard way to label AI involvement for transparency, including an unambiguous signal when no AI was used (cred-cr-no).

2. Format

cred-[human contribution]-[ai contribution] [version]

3. Contribution codes

All codes use only the letters c, r, e, d — easy to remember as "cred" (as in street cred, credits, credibility).

Code	Level	Meaning
no	0	No involvement — absent or only trivial tools (e.g., spell-check)
re	1	Reviewed — checked, approved, minor corrections
ed	2	Edited — substantially revised or rewrote
dr	3	Directed — guided, prompted, shaped the work
cr	4	Created — produced the original content

The scale reflects intensity of contribution: no < re < ed < dr < cr

4. Common examples

Badge colors indicate primary authorship at a glance: green when the human created the core content, red when AI created it, and yellow for balanced collaboration.

Tag	Human	AI	Typical Scenario
	Created	None	Fully human-authored, no AI involved
	Created	Reviewed	Human wrote it, AI contributed minor changes
	Created	Edited	Human wrote it, AI polished style
	Created	Directed	AI guided structure, human produced content
	Created	Created	Iterative collaboration, balanced contribution
	Edited	Created	AI drafted, human substantially rewrote
	Directed	Created	Human prompted and curated, AI generated
	Reviewed	Created	AI generated, human approved with minimal changes
	None	Created	Fully AI-generated

5. Attribution layers

Short form — minimal inline reference:

cred-dr-cr 1.0

Standard form — includes tool:

cred-dr-cr 1.0 | Claude Opus 4.5

Extended form — full disclosure statement:

cred-dr-cr 1.0 | Claude Opus 4.5 (Anthropic) | February 2026

The author provided detailed prompts, source materials, and structural
guidance. AI generated the initial draft. The author reviewed and
curated the output, making selective edits for accuracy and tone.
The author takes full responsibility for the final content.

6. Usage guidelines

1. Capture — choose codes that honestly reflect the work process
2. Refine — attribute distinct parts separately (e.g. Text: cred-cr-ed 1.0 | Images: cred-dr-cr 1.0)
3. Evolve — update attribution as works and contributions change
4. Disclose — reveal more for complex or uncertain cases, using the extended form

7. Responsibility

The cred standard describes process, not accountability. The human contributor always bears responsibility for the final work's accuracy, ethics, and fitness for purpose — regardless of AI involvement level.

8. Related initiatives

Several initiatives are currently working toward establishing transparent, interoperable standards for AI content attribution and labeling. These efforts aim to help creators, developers, and users identify the origins and authorship of AI-generated content across platforms and media.

• AI Attribution — An initiative exploring frameworks and metadata conventions to enable reliable attribution of AI models, datasets, and outputs.
• AI Labels — A collaborative project developing open standards for labeling AI-generated or AI-assisted content, to promote transparency and trust in digital media.
• C2PA (Coalition for Content Provenance and Authenticity) — An industry consortium defining a technical standard for certifying the provenance of digital content, including AI-generated assets.
• Content Authenticity Initiative (CAI) — A community-led effort to provide tools and standards for verifying content authenticity and attribution.
• AID Framework (University of Waterloo) — A guide outlining principles for transparent and responsible use of AI tools in research and creative work, emphasizing disclosure, integrity, and accountability.
• AACC Framework — A proposed framework for labeling creative works that incorporate AI, offering practical guidelines for writers, artists, and publishers to communicate how AI was used in their process.

These projects and frameworks share similar goals with cred, focusing on verifiable provenance, standardized metadata, and ethical transparency in AI systems. Including references to them helps situate cred within the broader ecosystem of AI attribution and content authenticity standards.

9. License

The cred standard is released under CC0 1.0 Universal (Public Domain). You may freely use, adapt, and build upon this standard without restriction.