AI Companies Destroy Millions of Books for Training
AI developers are anonymously buying physical books in bulk, scanning them for training data, then destroying the copies. This practice, likened to 'Fahrenheit 451,' is erasing rare books while boosting booksellers' profits. Legal battles over fair use continue, with Elon Musk advocating preservation.
Quick Take
AI firms are scanning and destroying physical books for training data, sparking outrage.
Booksellers report surge in sales but fear loss of rare and out-of-print titles.
Court rulings may allow destructive scanning under fair use, while separate case resulted in $1.5B settlement.
Market Impact Analysis
NeutralNo direct crypto market impact; article concerns AI book acquisition, not cryptocurrency.
Speculation Analysis
Key Takeaways
- AI developers are mass-purchasing physical books, scanning them for training data, and destroying the copies, permanently erasing rare titles.
- Booksellers report a sales surge from roughly 20 to several hundred books weekly, but fear the loss of out-of-print works.
- A federal judge ruled destructive scanning of legally purchased books as fair use, while Anthropic paid $1.5B for using pirated books.
- Demand concentrates on pre-2023 books to avoid AI-generated "slop," with Elon Musk publicly opposing the practice.
What Happened
AI companies, through anonymous intermediaries, are buying physical books in bulk, scanning them for training data, and then trashing the copies. The practice—likened to a dystopian novel—is alarming booksellers and preservationists as rare and out-of-print titles disappear. One bookseller saw weekly orders jump from 20 to several hundred books, driven by AI firms seeking high-quality text. The buyers insist on confidentiality, reflecting the legal and ethical sensitivity of the operation.
The Numbers
A single bookseller’s weekly sales rocketed from around 20 books to several hundred. A San Francisco federal judge deemed destructive scanning of legally purchased books as fair use, clearing a legal path. In a related case, Anthropic agreed to a $1.5 billion settlement for using pirated books, paying authors roughly $3,000 per title. The hottest commodities are books published before 2023—before the web was flooded with AI-generated content.
Why It Happened
AI developers desperately need high-quality, human-only text to train models. As AI-generated "slop" saturates the internet, pre-2023 books offer a pristine data source. Legal ambiguity has greased the wheels: courts have ruled that buying and then destroying books for scanning is fair use. The race for better models incentivizes companies to grab any text they can, regardless of the collateral damage to physical archives.
Broader Impact
This trend could wipe out rare and out-of-print books forever, eroding cultural heritage. It sets a dangerous precedent where physical source materials are consumed and destroyed for digital gain, potentially undermining libraries and collector markets. Authors may see more compensation through settlements, but the permanent loss of unique texts is irreversible.
What to Watch Next
- Pending copyright cases could overturn the fair use defense for destructive scanning, forcing companies to preserve books.
- Public pressure, including Elon Musk's call to save rare books, may push firms toward non-destructive digitization.
- The used-book market may face price spikes and scarcity as AI demand competes with collectors.
This article is for informational purposes only and does not constitute financial advice.
Always late to trends?
Join for the latest news, insights & more.
Disclaimer: Bytewit is an independent media outlet that delivers news, research, and data.
© 2026 Bytewit. All Rights Reserved. This article is for informational purposes only.
