Transparency over perfection. If you made it with AI, show your work.
AI collaboration is here. Transparency makes it legitimate.
Disclosure, not judgment. Like a nutrition label — informative, not prescriptive.
Regardless of AI percentage, a human signs off. That's the point.
Tell your AI: "Let's use Provenance Label standards. Track our collaboration."
Write, create, build. The AI tracks contribution throughout the session.
When done, ask: "Generate the Provenance Label." Review, approve, or adjust.
Bottom of posts, READMEs, artist statements — wherever fits your medium.
Copy this prompt into your AI at the end of any collaborative session. It calculates percentages and formats the label — you review and approve.
Based on our collaboration in this session, generate a Provenance Label v1.0. @provenance 1.0 author: [my name] date: [today YYYY-MM-DD] human: [X] ai: [Y] tools: [your name and version] note: [one plain sentence describing how we worked together] Requirements: - human + ai must equal 100 - Calculate based on: who initiated content, who made decisions, who edited, what percentage of final output came from each - Be honest, not flattering - Output only the label block above, ready to copy-paste
Uses PLGen syntax — the open format both short and long outputs are generated from.
The same label, two formats depending on where you're placing it:
PL v1.0 | Shelton Davis | 2026-02-17 | Human 65% · AI 35% · Claude Sonnet 4.6Use in: social posts, bylines, inline footers, Substack notes
─────────────────────────────────────
PROVENANCE LABEL v1.0
─────────────────────────────────────
Author: Shelton Davis
Date: 2026-02-17
Human: 65%
AI: 35%
Tools: Claude Sonnet 4.6
Note: Concept and direction human.
Draft and structure AI.
Edited for voice.
─────────────────────────────────────
provenance-label.org/specUse in: blog posts, articles, documentation, research papers
No. It assumes AI collaboration can be valuable when disclosed honestly.
It tracks who initiated content, who made decisions, who edited, and what percentage of final output came from each contributor. It's an estimate — your approval is the final step.
No. Aim for honest, not perfect.
Use the prompt above. Most modern AI tools can track collaboration if asked. For existing work, estimate retroactively.
Yes. It's open. If you change it significantly, version it (v1.1, v2.0). See the Spec for the full syntax rules.
Yes, like any self-disclosure system. It relies on creator integrity. Your reputation is on the line.
No. This is disclosure, not verification. Think nutrition label — it tells you what's in it, not whether the ingredients are ethically sourced.
Created by Shelton Davis (Helper-ID, Empathy Lab) in February 2026. Born from a blog collaboration that needed better disclosure. Open standard. Free forever.
GitHub: github.com/provenance-label
License: CC BY 4.0 (standard) / MIT (code)
Contact: shelton@theempathylab.com