A company that had promoted a service to scan and destroy books for use as AI training data has removed references to it from its website. It now says the service was never launched, following criticism over the prospect of physical books being destroyed to support AI data collection.

The company’s earlier claims suggested that books could be digitised and then disposed of after scanning, raising concerns about copyright, preservation and the treatment of printed works. The article does not provide further details on how many books were involved, whether any scanning took place, or the company’s identity.