AI Use Policy for 3rd Parties

Policy on the Use of AI Training in for Third Parties

Effective date: 02/05/2026 

1. Purpose

This page explains how third parties may (and may not) use Polaris QCI Publishing content in AI-enabled systems. It is intended for AI developers, platform providers, libraries, aggregators, and other organizations seeking permission to use our books or related materials.

2. Core Policy

Polaris QCI Publishing does not authorize the use of our books, chapters, figures, or other content for training, fine-tuning, or otherwise improving public-facing foundation models, except under an express written license. We may consider requests for licensed search and retrieval uses (including retrieval-augmented generation) that provide attribution, limit excerpts, and include reasonable safeguards against reconstruction of our works.

3. Definitions

3.1 Public-facing foundation model training
Using our content (in whole or in part) to pretrain, continue pretraining, fine-tune, instruct-tune, or otherwise improve a broadly deployed general-purpose model by modifying model weights or parameters.

3.2 Retrieval-augmented generation (RAG)
A system design in which a model answers a user query by retrieving relevant passages from a licensed corpus at query time and using those passages as context. In a RAG design, the corpus supports retrieval and grounding; it is not used to train or improve the model itself.

4. What Is Not Authorized Without a Separate Written License

Unless Polaris QCI Publishing grants explicit written permission, third parties may not:

• Use Polaris content to train, fine-tune, or otherwise improve any public-facing foundation model or general-purpose model.
• Use Polaris content (or derived representations, such as embeddings or indexes) in any workflow that results in persistent changes to a model’s weights or behavior attributable to exposure to our content.
• Build or distribute datasets that contain Polaris content for AI training, including scraped, scanned, or extracted text or images.
• Enable outputs that reproduce or reconstruct substantial portions of a work (including through iterative prompting, bulk export, or systematic extraction).
• Provide full-chapter translation, chapter-level paraphrase, or “book-equivalent” outputs that substitute for lawful access to the published work.

5. What We May Consider Licensing

We may consider licensing arrangements for controlled retrieval and discovery use cases, including:

• RAG-based question answering over a licensed corpus, with appropriate access controls
• Search, indexing, and snippet display for discovery and navigation
• Citation-enabled summarization that points users to lawful access (purchase, library access, or authorized platform access)

Any permitted use will be governed by a written agreement defining scope, safeguards, attribution requirements, excerpt limits, security standards, and termination/deletion obligations.

6. Typical Safeguards We Expect (RAG and Search Uses)

If we consider a license, we generally expect measures such as:

• Attribution and citation to the relevant title/section where feasible
• Excerpt limits (e.g., snippet-length quotations, not extended passages)
• Anti-reconstruction controls, including rate limiting and monitoring to prevent systematic extraction
• Access control for the licensed corpus (authenticated users, contractual restrictions, no open bulk download)
• Security controls appropriate to the content and deployment context
• Clear separation from model training, including contractual restrictions and technical assurance

7. Updates

We may update this policy as technology, licensing practices, and legal frameworks evolve. Any permissions granted are governed by the written agreement in effect at the time of licensing.