We live in a moment when the question “did a human write this?” has become genuinely hard to answer, and genuinely important.
HumanProof Registry exists to answer it honestly.
We don’t claim AI tools don’t exist, or that good writers never use them. What we do recognize is the difference between work that is human-directed, shaped by a person’s ideas, voice, judgment, and responsibility, and content that is generated, lightly edited, and published at scale.
That difference still matters. We’re here to make it visible.
What We Recognize
HumanProof maintains a public directory of works and authors that meet our standard of human-directed authorship. This includes:
- Independent blogs and personal essays
- Newsletters and Substack publications
- Books and long-form journalism
- Social media accounts with a genuine human voice
- Reports, research, and policy writing
Works in our directory carry one of two designations:
Recognized: Editorially selected by our team, or approved following self-nomination. Based on qualitative assessment of the work: the ideas, voice, and editorial decisions are clearly human in origin.
Certified: A formal, evidence-based verification. The author demonstrates their process through drafts, version history, and editorial notes. Certification is available as a paid service and carries full documentation suitable for professional or legal contexts.
Our Standard
A HumanProof work is one where a human made the creative and editorial decisions: the ideas, the voice, the judgment, the final approval. Tools may assist; humans are responsible.
We don’t use algorithmic AI detectors, which are unreliable and easy to game. Recognized works are assessed editorially. Certified works are reviewed by independent paid human peers against real evidence of process.
Who We Are
HumanProof Registry is incubated by the Media and Journalism Research Center (MJRC), a member of the Coalition for Content Provenance and Authenticity (C2PA). Our verification process is fully independent — assessments are carried out by peer reviewers, not by MJRC or submitting authors.
All evidence submitted for certification, draft files, version histories, metadata, is handled securely and deleted within 30 days of a decision unless the author opts for longer retention.