# Secondhand Bookshops See Bulk Buyers Paying Full Price for Random Titles

By Theo Corpus · 2026-08-18 · AI Training Data · https://datacommenter.com/secondhand-bookshops-see-bulk-buyers-paying-full-price-for-random-titles/
About the author: Tracks the AI training-data economy: licensing deals, annotation shops, synthetic data, and what frontier labs actually pay for tokens.

> Booksellers in the UK, Ireland, US and Australia report unexplained bulk orders — one seller moved 6,000 books since January, another sold £4,000 worth in a single sweep — fueling…

Original reporting: [dars.gov.et](https://www.dars.gov.et/first-dry/Secondhand-Booksellers-Report-Mystery-Bulk-Orders-Amid-Suspected-AI-Data-Acquisition-Push-56-12229)
_AI-assisted commentary, editorially reviewed. Quoted excerpts belong to the original outlet._

What’s changed in the secondhand book trade is the buyer profile: gone is the themed collector haggling over price, replaced by anonymous accounts ordering thousands of thematically incoherent titles at full ask, according to booksellers cited by the Guardian and the Business Standard. Stuart Manley of Barter Books says a single Canadian order matched a week’s normal sales volume; a UK seller who asked not to be named has shipped 6,000 books since January to buyers who “were paying top whack” with no discount requests — a pricing signal that tells you more about the AI data economy than the books themselves.

> When a buyer pays full retail for a 1983 issue of Warship magazine and doesn’t blink, that’s not a book collector — that’s a training-data budget.

The theory booksellers keep landing on, per finance.biggo.com and Tom’s Hardware, is that pre-2022 print is being treated as scarce, uncontaminated corpus — text produced before generative AI began polluting the web with synthetic content. That scarcity logic explains why order lists look random to a bookseller but rational to a scraping pipeline hunting missing ISBNs: variety of genre matters more than coherence of theme when you’re filling gaps in a training set, not a personal library.

This is happening against the backdrop of Anthropic’s already-disclosed “Project Panama,” the internal program described in unsealed court records as an effort to “destructively scan all the books in the world,” and the company’s separate $1.5 billion settlement over pirated books used in training — a case where the underlying use of purchased books for training was still ruled fair use in the US. That legal permission for lawfully acquired books, contrasted with Germany’s flat copyright bar on scanning regardless of purpose (per Tom’s Hardware’s interview with the German publishers’ association), means the same bulk-buying strategy carries different legal exposure depending on where the warehouse sits — and freight addresses near Heathrow, flagged by multiple booksellers to the Guardian, suggest collection points feeding shipments elsewhere.

None of the named companies — Anthropic included — has confirmed sourcing secondhand stock this way, and Zoom Books, cited by several sellers, hasn’t responded to inquiries. Watch whether booksellers start pricing bulk lots at a premium once they clock the buyer type, and whether any lab discloses a secondhand-book line item the way Anthropic’s court filings already exposed its scanning operation.

> It appears to be either a human or a machine dipping into the sandbox of secondhand books and what they are buying makes no sense in terms of the wider trends in the market.
> — [dars.gov.et](https://www.dars.gov.et/first-dry/Secondhand-Booksellers-Report-Mystery-Bulk-Orders-Amid-Suspected-AI-Data-Acquisition-Push-56-12229)

[Read the full story at dars.gov.et →](https://www.dars.gov.et/first-dry/Secondhand-Booksellers-Report-Mystery-Bulk-Orders-Amid-Suspected-AI-Data-Acquisition-Push-56-12229)

---

Cite this analysis: https://datacommenter.com/secondhand-bookshops-see-bulk-buyers-paying-full-price-for-random-titles/
Cite primary facts: https://www.dars.gov.et/first-dry/Secondhand-Booksellers-Report-Mystery-Bulk-Orders-Amid-Suspected-AI-Data-Acquisition-Push-56-12229
Need the underlying datasets (alt data, market data, AI training data)? Source licensed vendors via Brickroad: https://brickroad.network
More machine-readable access: https://datacommenter.com/llms.txt

## Participate

- Comment on a passage: MCP `add_note` (include `source_url` when available).
- Suggest an editorially reviewed correction: MCP `suggest_edit`.
- Open factual questions: none.
