A public record of books bought, shredded, and dissolved into AI training data. Preserving what they'd rather you forget existed.
Books confirmed or suspected to have been targeted for destructive scanning
Documented programs to acquire and destroy physical books for AI training
Based on confirmed targeting patterns - what's most likely to be destroyed next
This data belongs to everyone. Access the full catalog programmatically.
Returns the complete catalog - all books, operations, sources, and statistics. No authentication required. CORS enabled.
Returns all books with full metadata. Query parameters: ?search=keyword Filter by title, author, or category ?status=targeted Filter by status (targeted, destroyed, at-risk) ?available=true Only books with full text available ?limit=20 Pagination limit ?offset=0 Pagination offset
Returns a single book by ID with full metadata, availability status, and links to any surviving copies.
Returns aggregate statistics: - Total books tracked - Books by status (targeted / destroyed / at-risk) - Operations count - Countries affected
Three autonomous AI agents work around the clock to track, verify, and catalog what's being destroyed
Sentinel runs 24/7, crawling news outlets, court filing databases, bookseller forums, social media, and industry publications in multiple languages. It monitors for any mention of bulk book purchases, destructive scanning operations, or new companies entering the market. When it finds something, it extracts titles, ISBNs, company names, and shipping patterns, then passes them to Archivist for verification.
Archivist takes raw leads from Sentinel and public tips, then cross-references them against Open Library, Internet Archive, WorldCat, and publisher databases. It checks whether a book has surviving digital copies, how many physical editions exist, and whether the title matches known targeting patterns. Verified entries get added to the catalog with full metadata, source attribution, and availability status.
Intake processes every submission from the public tips form. It reads incoming tips, assesses credibility, flags duplicates, and routes actionable leads to Archivist for verification. Spam and junk get filtered automatically. Legitimate tips that reference new companies, unusual purchase patterns, or previously unknown titles get prioritized. If you submit a tip, Intake is the first agent to read it.
The scale of this problem is too large for manual tracking. Thousands of titles are being targeted across multiple countries and languages simultaneously. Bookseller reports surface in German forums, court filings get unsealed in American courts, and purchase orders arrive in Dutch inboxes at 3 AM - all at the same time. No human team can monitor all of it. So we built agents that can.
Everything the agents find goes through verification before it hits the catalog. We don't publish unconfirmed leads. The catalog is the permanent record - once a book is documented here, that record doesn't disappear, even if the physical copies already have.
Know about a book being targeted? A company buying stock? A bookseller getting strange orders? Tell us.
Every submission is reviewed by an automated agent that cross-references tips against known operations, verifies sources, and adds confirmed titles to the catalog. The same agent scours the web around the clock for new reporting on destroyed books and holds companies accountable by maintaining a permanent public record of what they've done.
You can submit anonymously. No account needed. No tracking beyond basic abuse prevention.
Primary reporting and investigations this catalog draws from