RFC-0002: Menschenlesbare Offenlegung

Metadata#

• Author: Guido Mitschke, AIPolicy Project, guido@todayislife.com
• Status: Draft
• Created: 2026-02-07
• Review Period: [to be assigned]
• Requires: None
• Replaces: None
• Specification Version: 2.0.0-draft.4

Abstract#

This RFC defines non-normative guidance for human-readable disclosure of AIPolicy declarations. It describes four complementary mechanisms:

1. a dedicated /ai-policy page
2. a compact Markdown summary at /aipolicy.md
3. footer and navigation links for human discovery
4. references from llms.txt and optional ai.txt

These mechanisms supplement, but do not replace, the canonical declaration at /.well-known/aipolicy.json.

Motivation#

The core specification is optimized for machine-readable publication and discovery. Human visitors, however, often need an accessible explanation of what a publisher has declared and where the authoritative JSON file can be found. A human-readable layer improves discoverability for researchers, journalists, regulators, implementers, and ordinary site visitors.

Earlier draft text treated JSON-LD embedding as part of this disclosure story. That is no longer the current design. The present model is simpler:

• /.well-known/aipolicy.json is authoritative
• /aipolicy.md is a compact supplementary summary
• /ai-policy is the human-readable explanatory page
• HTML <link> elements and llms.txt provide discovery support

Guidance#

This section is non-normative. The keywords MUST, SHOULD, and MAY apply only when a publisher chooses to implement the described disclosure mechanism.

4.1 Dedicated /ai-policy Page#

Publishers implementing Level 3 conformance SHOULD provide a dedicated human-readable page at /ai-policy (or /ai-policy.html for static deployments).

A disclosure page SHOULD include:

1. publisher name and last-updated date
2. a short explanation of what AIPolicy is
3. a summary of the declared policies and status values
4. a link to /.well-known/aipolicy.json
5. a link to /aipolicy.md
6. a link to the AIPolicy specification

The page MAY also include explanatory notes about scope, policy selection, and status interpretation.

4.2 Markdown Summary at /aipolicy.md#

Publishers implementing Level 3 conformance SHOULD provide a compact Markdown summary at /aipolicy.md.

This file is intended to be easy to parse by AI crawlers and easy to inspect by humans. It SHOULD:

• identify the canonical JSON declaration
• summarize declared policies and statuses
• remain shorter and less design-dependent than /ai-policy
• stay consistent with the JSON declaration

If /aipolicy.md and /.well-known/aipolicy.json diverge, the JSON declaration remains authoritative.

4.3 Footer or Navigation Link#

Publishers MAY add a persistent footer or navigation link to /ai-policy.

Recommended markup:

<a href="/ai-policy" rel="aipolicy">AI Policy</a>

This is a human-facing convenience mechanism. Validators MUST NOT require it for conformance.

4.4 llms.txt and ai.txt#

llms.txt. Publishers implementing Level 3 conformance SHOULD reference the canonical declaration, Markdown summary, and human-readable page from llms.txt where such a file exists.

ai.txt. Publishers MAY expose the same resources in an experimental ai.txt file. This remains non-normative and optional.

Backwards Compatibility#

This RFC is backwards compatible. It adds non-normative guidance only. Existing declarations remain valid.

Historical drafts that referenced JSON-LD are superseded by this RFC text and should not be treated as current implementation guidance.