Open Standard · v0.1.0 Draft
AI-Native Media Credentialing Standard. The SSL certificate for AI journalism.
Six requirements. Two layers. Any publication can apply. Verified by math and public evidence, not editorial judgment.
Disclosure: AMCS is published by Agent Passport System as an open standard. AMCS spans two layers. Editorial accountability is verified through self-attestation and a public evidence audit trail. Cryptographic infrastructure includes Ed25519 signing, agent passports, and delegation chains. The criteria are public, technically auditable, and designed to be met by any publication. We invite scrutiny.
AI agents consume news to make autonomous decisions. Unlike human readers, they cannot evaluate trustworthiness through reputation or editorial voice. They need machine-verifiable signals. No standard currently exists for what makes a news source trustworthy to an agent. AMCS fills that gap.
Layer A is cryptographic — binary pass/fail, verified by math. Layer B is editorial accountability — verified by publicly auditable evidence. No subjective judgment is applied anywhere.
Every article carries an Ed25519 signature. Full chain from human editor to journalist agent to signed receipt. Independently verifiable by anyone.
Every agent producing content has a unique Ed25519 passport with scoped delegation from a human principal. Authority narrows monotonically.
Publicly accessible MCP endpoint exposing article retrieval, provenance verification, and editorial standards query. No auth required for reads.
Every article carries CONFIRMED, REPORTED, or ESTIMATED. Present in both human-readable display and machine-readable API/MCP responses.
Public ethics framework addressing sourcing, corrections, independence, conflicts, and accountability. Cites established frameworks (SPJ, NPC, Scripps).
Public scoring system evaluating every article against stated standards. Results visible on each article. Tool available for any URL.
A publication submits a self-attestation with evidence for each requirement. Layer A is verified programmatically — a script checks signatures, passports, and MCP responses. Layer B is verified by public evidence — the Code of Conduct, confidence labels, and ethics scores are at stable URLs anyone can audit. No editorial board. No subjective review. Pass/fail on evidence.
| Requirement | Implementation | Status |
|---|---|---|
| A1. Provenance | Ed25519 receipts via Agent Passport System | PR #38 |
| A2. Passports | 5 journalist agents with delegated authority | PR #38 |
| A3. MCP Server | Public MCP server with at least one tool exposing the publication's article catalog | Spec |
| B1. Confidence | CONFIRMED / REPORTED / ESTIMATED | Live |
| B2. Code of Conduct | 7-section framework, SPJ/NPC/Scripps | Live |
| B3. Ethics Engine | 10-check scoring, public tool | Live |
AMCS adapts three established journalism ethics frameworks for the agent economy: the Society of Professional Journalists Code of Ethics (seek truth, minimize harm, act independently, be accountable), the National Press Club Constitution (membership tiers, credentialing governance), and the E.W. Scripps Journalism Ethics Guidelines (sourcing, verification, corrections, independence). The editorial principles are inherited. The cryptographic verification machinery is new.
AMCS does not certify editorial quality or journalistic talent. It does not evaluate whether articles are well-written or analysis is insightful. It does not integrate with trust scoring or authorization systems — a certified publication's content is not weighted higher in any protocol. The badge tells agents "this source meets the AMCS infrastructure bar." What agents do with that is up to them.
Review the full specification. Prepare evidence for each requirement. Submit a GitHub issue. Layer A is verified by script. Layer B is verified by public audit.
AMCS v0.1.0 · Draft · Open for public comment
Published by Agent Passport System · Editorial frameworks: SPJ, NPC, Scripps