AI firms are buying up old books because they are the last slop-free data left

The Next WebPublished Jul 22, 2026

ISBNdb supplies pre-2022 printed books to AI labs for training data to avoid AI-generated text contamination and data poisoning. The process involves destroying books by removing spines for scanning, with confidentiality agreements to keep buyers anonymous.

#ai-training-data #data-poisoning #isbndb #model-collapse #pre-2022-books

AI firms are buying up old books because they are the last slop-free data left

More in Machine Learning

Arlequin AI raises €28M to build novel AI models that learn complex relationships at scale

DeepSeek launches V4.1-Flash and retires V4-Pro, its flagship model

Nvidia and Palantir fine-tune a 30B Nemotron model for Nvidia’s supply chain. It beats a model 18 times its size.