robots.txt told crawlers what to do.
ai-transparency.json tells users what AI did.
The Deployer's answer to Article 50 — in 5 minutes.
The EU AI Act Article 50 demands machine-readable AI content disclosure.
Existing standards weren't built for this.
Inbound focus only. Tells scrapers what NOT to train on. Doesn't solve the Article 50 outbound disclosure obligation.
❌ Wrong directionCryptographic overkill. Requires certificate infrastructure. Designed for images and video — not AI-generated text content.
❌ Too complexMassively overloaded. Demands model training details and data labeling. Unusable for normal web publishers.
❌ Impractical
A single manifest at /.well-known/ai-transparency.json that fulfills
your Article 50 obligations — no crypto, no certificates, no enterprise tooling.
Create one JSON file, upload it. That's it. No infrastructure, no certificates, no vendor lock-in.
Chatbot (ATS-4) vs. Blog (ATS-2) vs. Legal pages (ATS-0). Don't declare your whole site as AI-generated.
Machine-readable disclosure that EU audit tools and compliance scanners can validate automatically.
Granular scope prevents search engines from penalizing human content because of an AI chatbot on the same domain.
JSON Schema reference enables automated compliance checking and continuous monitoring.
Built-in accessibility. Transparency that isn't accessible to everyone isn't transparent at all.
Five tiers that honestly declare who drafted the content — human, AI, or both.
No AI involvement whatsoever. Pure human authorship.
Human-authored with AI assistance for spelling, grammar, or formatting.
AI generates fragments and sections. Human assembles, edits, and curates.
AI generates the full draft. Human reviews, validates, and approves.
Fully AI-generated content with minimal or no human oversight.
Create /.well-known/ai-transparency.json on your server. Done.
{
"$schema": "https://ai-transparency-protocol.org/schema/v2.0.json",
"eu_ai_act": {
"article_50_compliant": true,
"effective_date": "2026-08-02",
"disclosure_method": ["http-header", "client-widget", "manifest"]
},
"policies": [
{
"scope": "/chatbot/*",
"ats_tier": "ATS-4",
"ats_extent": "E4",
"human_oversight": false
},
{
"scope": "/blog/*",
"ats_tier": "ATS-2",
"ats_extent": "E2",
"human_oversight": true
},
{
"scope": ["/impressum", "/agb", "/datenschutz"],
"ats_tier": "ATS-0",
"description": "Strictly human-authored, no AI disclosure required."
}
],
"transparency_policy_url": "https://example.com/ai-transparency",
"contact": "compliance@example.com",
"generator": "TrueSource AI Middleware"
}HTTP Headers + JSON Manifest + HTML <link> auto-discovery
<link rel="ai-transparency" href="/.well-known/ai-transparency.json">
Structured HTML comment block — visible in "View Source" and DevTools
<!-- ATP: ATS-2 · Co-Creative · Art. 50 compliant -->
HTTP response headers for automated compliance scanners
AI-Transparency: article-50-compliant
How we stack up against existing standards.
| Feature | AI Transparency Protocol | C2PA | IETF OETP | llms.txt |
|---|---|---|---|---|
| Text content support | ✓ | ✗ | ✓ | ✗ |
| Per-route scope | ✓ | ✗ | ✗ | ✗ |
| ATS tier integration | ✓ | ✗ | ◐ | ✗ |
| No crypto/certificates needed | ✓ | ✗ | ✓ | ✓ |
| Article 50 compliant | ✓ | ✓ | ◐ | ✗ |
| 5-minute deployment | ✓ | ✗ | ✗ | ✓ |
| SEO-safe | ✓ | N/A | ✗ | N/A |
| JSON Schema validation | ✓ | N/A | ✓ | ✗ |
| Editorial Exemption proof Art. 50 Abs. 5 |
✓ | ✗ | ✗ | ✗ |
The criticism: "It's not an official EU standard."
The history: Neither was anything in /.well-known/ — until it was.
| Standard | Official at Launch? | Status Today |
|---|---|---|
robots.txt |
❌ Private proposal (1994) | ✅ De-facto universal standard |
security.txt |
❌ Indie project | ✅ RFC 9116 (IETF Standard) |
ads.txt |
❌ IAB initiative | ✅ Industry standard |
humans.txt |
❌ Community project | ⚡ Widely adopted |
ai-transparency.json |
❌ Open Source (2026) | ⏳ Building momentum |
Every /.well-known/ standard began exactly where ATP is now.
Adoption creates standards. Standards don't create adoption.
Six principles that guide every design decision in this protocol.
Users have a fundamental right to know when they consume AI-generated content. This is not just law — it is an ethical imperative.
If a standard requires crypto infrastructure, only the biggest companies will adopt it. The other 99% of the web stays opaque. We choose pragmatism.
Calling an entire website "AI-generated" because of one chatbot is inaccurate and harmful. Per-route policies enable truthful disclosure.
Compliance tools should not be proprietary moats. MIT license, open spec, vendor-neutral. Small businesses need free infrastructure.
Transparency that isn't accessible to people with disabilities is not transparent. WCAG 2.1 AA is a requirement, not a feature.
Self-declaration without enforcement is meaningless. That's why ATP includes JSON Schema validation and automated compliance scanning — the manifest declares, the scanner verifies.
Two open protocols that solve the entire AI web integration challenge.
€15M fines for missing AI disclosure
ai-transparency.json manifest + HTTP headers + client widget
35% of searches go to AI — your brand is invisible
Emotional + contextual triggers in Schema.org for LLM optimization
View VibeTags →
The clock is ticking. Article 50 enforcement starts in --- days.
Start with one JSON file. Protect your business.