Amazon destroys rare books to train AI models: Ethical debate threatens cultural heritage

Rare books are those not available in digital form online, containing unique content valuable for AI training, such as old manuscripts and first editions.

Because large language models need diverse, non-online data to improve performance, and rare books provide unique content not available from digital sources.

Destroying rare books could lead to loss of valuable Arabic and Islamic manuscripts, threatening the region's cultural heritage and historical knowledge.

High-quality digitization and preserving originals, plus collaboration with libraries and archives, can leverage content without destruction.