The bottom line: Bulk purchases of antiquarian books by a Canadian merchant suggest destructive scanning for AI training and raise regulatory compliance concerns.
A Canadian antiquarian bookseller is systematically acquiring large quantities of books in Germany, possibly for AI training material — and destroying them afterwards. This raises questions about copyright compliance and the practical implementation of the EU AI Act.
Based on research, a Canadian antiquarian bookseller is systematically and extensively purchasing books from German antiquarian bookstores. The availability and pattern of these purchases suggest that the acquired copies are being used for AI training material and subsequently destroyed — rather than permanently returning to commerce or being preserved in archives.
For Chief Data Officers and compliance officers, this is relevant because it exposes the practical gray zone between legal data sourcing and copyright infringement. The destructive scanning of books for large language models potentially violates German copyright law and provisions of the EU AI Act, which requires transparent and lawful training data. Particularly for high-risk AI systems, traceable sourcing and licensing of training material is required.
The practice also endangers German cultural heritage: unique or rare copies are lost that would otherwise be available for research and collection management. For organizations that train or deploy AI models, this also raises the question of how transparent and legally compliant their own supply chain for training data must be — especially when external data providers or intermediaries are involved.
Source: www.golem.de · Published June 24, 2026
Lumi AI News — AI-assisted curation pursuant to Article 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.1.