AI Companies Bulk-Buying and Shredding Rare Books for Training Data
July 27, 2026
AI firms are allegedly using anonymous services like ISBNdb to acquire and scan rare books, destroying the physical originals to facilitate training. Companies are prioritizing pre-2022 texts to avoid training on synthetic data, sometimes using NDAs to mask the process.
HOW THIS AFFECTS YOU
●
researcherThe availability of high-quality, non-synthetic datasets may depend on these aggressive acquisition tactics.
●
policyThis raises significant questions regarding intellectual property and the preservation of cultural heritage.