Skip to content
Ars Technica·

📖AI Companies Destroying Books to Train Models

Rare booksellers are on high alert as AI firms bulk order and destroy print copies

TL;DR

AI companies are reportedly destroying rare books to train their models, raising alarms among sellers. The Internet Archive uses non-destructive methods for preservation, while some firms like OpenAI pledge to protect rare texts.

AI companies are tearing through rare book collections to train their models, a practice that's likely more widespread than initially reported. This is a serious issue for anyone involved in the preservation of historical and literary heritage. The Internet Archive has been using non-destructive methods since 2009, scanning over 3 million pages without damage. But now, rare booksellers are on high alert as suspicious bulk orders flood their shops. A lawsuit against Anthropic is expected to reveal more about this destructive trend.

AI Companies Destroying Books to Train Models — Ars Technica

Key Points

1

Google patented a non-destructive scanning tech in 2009 to preserve books without damage

2

The Internet Archive has scanned over 3 million pages using clean, dry human hands

3

A lawsuit against Anthropic is expected to expose more about AI's destructive practices

4

OpenAI and Microsoft are working with Harvard librarians on a project to train models on public-domain texts

5

Elon Musk’s xAI pledges not to destroy rare books for training its AI

Why It Matters

Rare book sellers face an ethical dilemma as large orders flood in. If you're selling historical or literary works, be wary of bulk orders that may lead to the destruction of unique collections. The Internet Archive's method ensures zero errors with proprietary software and human care.

rare-booksAI-training-datapreservationInternet-Archive

Frequently Asked Questions

Why does this matter?

Rare book sellers face an ethical dilemma as large orders flood in. If you're selling historical or literary works, be wary of bulk orders that may lead to the destruction of unique collections. The Internet Archive's method ensures zero errors with proprietary software and human care.

What happened?

AI companies are reportedly destroying rare books to train their models, raising alarms among sellers. The Internet Archive uses non-destructive methods for preservation, while some firms like OpenAI pledge to protect rare texts.

Comments

Subscribe to join the conversation...

Be the first to comment

Enjoyed this article?

Get it daily. 7am. Free. Reads in 5 minutes.

Join 2,945 builders reading daily.

Also get