Rights, sources, and responsible Book RSS publishing

A practical publishing checklist for public-domain books, licensed content, owned works, excerpts, catalog metadata, scans, OCR, cover images, AI summaries, and source terms.

This page provides general publishing guidance, not legal advice. Rights depend on the work, jurisdiction, contract, license, source terms, and intended use.

Assign a rights status before transforming content

Do not wait until the feed is ready to ask whether the material can be distributed. Every source record should have a documented status and provenance. The publishing build should enforce that status. AI extraction, OCR, format conversion, or access to a file does not create permission to republish it.

StatusTypical permitted feed behaviorRequired evidence
OwnedMetadata, summaries, authorized excerpts, files, and full text according to the owner’s policyOwnership or publishing agreement
LicensedOnly the uses, territories, formats, and term granted by the licenseLicense text and expiration or territory fields
Public domainMetadata, scans, OCR, and full text subject to source and privacy considerationsDocumented public-domain analysis for relevant jurisdiction
Open licenseUses allowed by the license with attribution, share-alike, or other conditionsExact license version and source attribution
Excerpt onlyLimited authorized excerpt plus canonical source linkPermission, contract, or a reviewed legal basis
Metadata onlyOriginal bibliographic summary and links; no protected full text or unauthorized coverSource terms and field-level provenance
Restricted / unknownNo public file or full text; hold for reviewResolution by the responsible owner or counsel

Keep source access separate from redistribution

A person may be allowed to read a purchased e-book, borrow a library file, or access a database without being allowed to republish its contents. An API may permit display of certain fields while restricting storage, image use, price caching, or derivative feeds. Record the source’s terms and required attribution. Build with supported APIs or exports rather than bypassing access controls.

Full text, scans, and OCR

Preserve the original source and mark OCR as a derived representation. Public-domain or licensed status should be tied to the specific edition or scan when necessary. A modern introduction, translation, annotation, cover, or typesetting may have different rights from the underlying historical work. Do not assume the entire package shares one status.

Scans can also contain private notes, addresses, medical information, or institutional records. Review sensitive collections even when the printed work is old. A full-text feed can amplify information far beyond its original context.

Covers, photos, and maps

A cover is not automatically reusable because it represents a book. Use publisher-supplied or licensed images under their display rules, public-domain images with provenance, or original branded cards. Do not substitute the cover of a different edition. For travel guides and city feeds, use authorized maps or original graphics and attribute sources when required.

Summaries, metadata, and AI

Write original summaries from approved facts. Do not copy long promotional descriptions or reviews without permission. An AI-generated paraphrase still needs grounded inputs, factual review, and a rights-aware workflow. Models should never decide that a work is public domain or that copying is fair use. The publishing system should rely on a documented status assigned outside the model.

Removal, correction, and expiration

Store license expiration dates and territory restrictions in machine-readable fields. When rights expire, remove files and full content from distribution promptly while preserving only the public metadata that may remain. Provide a clear contact path for rights questions and corrections. Do not expose internal legal correspondence or personal information on a public status page.

Pre-publication checklist

  • What exact work, edition, image, file, or excerpt is being published?
  • Who supplied it, and under what terms?
  • What jurisdictions, territories, formats, and dates apply?
  • Can the project store and transform the source, or only display selected fields?
  • Are attribution, link, price, or image requirements satisfied?
  • Does the item contain private or sensitive information?
  • Can the build prove the status and stop publication automatically when conditions fail?

For AI-specific controls, read AI and Agentic Book RSS. For scanned sources, use the review model in OCR Book Scans to RSS.