Overview
Elthian AI Licensing Policies define exactly how AI systems may interact with your content. Each policy in the suite is a standardized, self-contained set of rules covering core AI use cases:
- Training (ingestion for model improvement, dataset building, fine-tuning, evaluation)
- AI Search / Indexing (referential discovery with attribution, non-substitutive excerpts, direct links)
- RAG / Product Use (context injection, assistants, copilots, internal tools, agents)
- Embeddings (vector storage, similarity search, persistence)
- Derivative / Substitutive Outputs (summaries, paraphrases, reproductions that could replace visiting the source)
- Commercial Use (monetization, product features, advertising-supported deployment)
Policies are expressed through machine-readable signals (JSON, headers, robots.txt, llms.txt, sitemap) generated by the plugin. They reduce ambiguity for crawlers, platforms, and future compliance tools while preserving your copyright.
The Elthian License Suite
The suite forms a clear “ladder” of permissions — from maximum protection to maximum openness. Choose based on your content goals.
| License | AI Training | AI Search | Product / RAG Use | Derivative AI Output | Typical Use Case |
|---|---|---|---|---|---|
| Elthian-NoAI | ❌ Prohibited | ❌ Prohibited | ❌ Prohibited | ❌ Prohibited | Sensitive, proprietary, paid, or personal content |
| Elthian-NoTrain | ❌ Prohibited | ✅ Allowed (referential only) | ❌ Prohibited | ❌ Prohibited | Publishers who want discovery without ingestion |
| Elthian-UseOnly | ❌ Prohibited | ❌ Prohibited | ✅ Allowed (internal, human-in-the-loop) | ❌ Prohibited | Enterprise internal copilots, assistants, workflows |
| Elthian-NonCommercial | ⚠️ Restricted | ⚠️ Restricted | ⚠️ Restricted | ⚠️ Restricted | Academic, research, non-profit use |
| Elthian-Public | ✅ Allowed | ✅ Allowed | ✅ Allowed | ✅ Allowed | Open knowledge, public datasets |
| Elthian-Custom | Configurable | Configurable | Configurable | Configurable | Enterprises, agencies, mixed-policy sites |
All policies:
- Allow Human Access (normal browsing) and Incidental Technical Copying (transient caches for delivery).
- Prohibit circumvention (proxy laundering, query stitching, reassembly).
- Require Attribution where search/use is permitted (credit + direct link).
- Include strong anti-substitution clauses (outputs must not replace visiting the source).
For full legal text, see the canonical pages at elthian.com/licensing.
How Policies Are Resolved
The plugin follows this order to determine the effective policy for any URL:
- URL-specific override (if defined for the exact path or matching pattern) → highest priority.
- Global default policy (site-wide setting) → fallback.
- If no policy is set or signals fail → implicit NoAI (safest default; no rights granted for AI use).
Resolution happens dynamically on every request/output generation, ensuring consistency.
Key Capabilities Explained
- Training — Using content to influence model weights/parameters (pretrain, fine-tune, RLHF, benchmarking, dataset curation).
- AI Search / Indexing — Building indexes for query-based discovery with short, non-substitutive excerpts (capped length, citations required, directs to source).
- RAG / Product Use — Injecting content at inference time for generation/assistance (persistent stores, agents, copilots, internal tools).
- Embeddings — Vector/latent representations stored for retrieval (ephemeral = deleted after query; persistent = long-term index).
- Derivative / Substitutive Outputs — Any generated text/image that reproduces expressive elements or lets users skip the original.
Lite version supports basic global + URL overrides. Pro adds granular scopes (post type, taxonomy, regex patterns), signed snapshots, and audit trails.
Policy Outputs in Action
Once set, your chosen policy automatically populates:
- robots.txt → User-agent directives (e.g., block GPTBot, anthropic-ai on NoTrain sites)
- llms.txt → Human/AI-readable notice
- HTTP Headers →
X-AI-License,X-AI-Policy-Ref, etc. - JSON →
/.well-known/ai-policy.json(canonical machine-readable declaration) - Sitemap →
ai-license-sitemap.xml(only overridden URLs)
All channels reflect the same resolved rules — no contradictions.
Choosing the Right Policy
- Want maximum protection? → NoAI
- Want discovery in AI search without feeding models? → NoTrain
- Need internal AI tools but no public exposure? → UseOnly
- Building research/non-profit tools? → NonCommercial
- Sharing openly? → Public
- Need mixed rules across your site? → Custom + overrides
Start with the global setting in the plugin dashboard, then refine with overrides as needed.
Next Steps
- Configure your global policy in the plugin
- Add URL overrides for mixed sites
- Verify outputs (see Getting Started → Verifying Outputs)
- Explore Pro features for verification and advanced scoping