Across Europe, antiquarian booksellers report surge in unusually large orders from Canadian company Zoom Books
Melike Pala
21 July 2026•Update: 21 July 2026
BRUSSELS
- ‘It appears to be yet another example of AI companies using vast quantities of copyright-protected works to train their language models, without consent and without payment,” says leading German publishers association
- Copyright violation difficult to prove in court because of legal loopholes
A strange phenomenon is unfolding across Europe’s antiquarian book market, where decades-old books are being bought in bulk, shipped across the Atlantic and reportedly scanned and destroyed for artificial intelligence training.
This has raised fresh concerns about copyright and the future of physical books.
The German Publishers and Booksellers Association, Germany’s leading publishing industry association, has called the practice “a clear and serious violation of copyright.”
The orders come primarily from Canadian company Zoom Books, according to media reports in the US, Germany, and Switzerland.
Sellers say the company has focused on nonfiction and academic titles published in the 1970s, often buying dozens or even hundreds of volumes from individual dealers.
The books are reportedly sent to temporary warehouses in Germany before being shipped to Canada and the US.
According to The Washington Post, many are then subjected to “destructive scanning,” a process in which the bindings are removed, the pages are digitized using high-speed industrial scanners, and the remaining physical books are recycled.
Zoom Books has denied participating in the digitization and destruction of books for AI development.
Speaking to Anadolu, Thomas Koch, press spokesperson for the German Publishers and Booksellers Association, said the organization has been informed by its members about these reports.
“We don’t know the extent of the issue. It appears to be yet another example of AI companies using vast quantities of copyright-protected works to train their language models, without consent and without payment,” he added.
Legal loopholes
However, copyright violations are difficult to prove in court because of legal loopholes.
“To date, however, such violations have been difficult to prove because there are no transparency requirements that would mandate the citation of
According to Koch, the absence of disclosure obligations makes it nearly impossible for authors and publishers to determine whether their works have been incorporated into AI training datasets.
He also emphasized that legal treatment differs significantly across jurisdictions.
“The legal classification of such situations is governed by the law of the country in which the purchaser is based or in which the act is carried out,” he said.
“Under German law, scanning books—for whatever purpose—is not permitted,” he said.
Recent US rulings involving Anthropic and Meta have generally favored AI developers under the fair use doctrine, although legal experts have stressed that the decisions do not amount to blanket authorization for all uses of copyrighted material.
Within the European Union, copyright rules take a different approach through the Digital Single Market Directive, which permits text and data mining under certain conditions while allowing rights holders to opt out.
Call for stronger legislation
Koch said court decisions alone cannot resolve the growing conflict between the publishing industry and AI developers.
“We have long been calling for the relevant legislation to be tightened up. Numerous legal cases have now highlighted the lack of legal certainty in this area,” he said.
He pointed to ongoing litigation initiated by Penguin Random House in Germany against OpenAI over alleged copyright infringements involving the children’s book, Der kleine Drache Kokosnuss, as an example of publishers increasingly seeking judicial remedies
According to Koch, AI developers should be required to disclose the copyrighted works used to train their models, enabling authors and publishers to verify whether their material has been used and ensuring creators receive appropriate compensation.
Threat to writers and publishing industry
Beyond copyright concerns, Koch warned that unrestricted AI training poses an existential challenge to the economic foundations of professional publishing.
He argued that AI systems trained on copyrighted books are now capable of producing new books almost instantly, contributing to an influx of low-cost, AI-generated titles on digital marketplaces.
“AI is now capable of generating books at the touch of a button. Such mass-produced, AI-generated books already pose a major and growing problem,” Koch warned.
According to Koch, many AI-generated publications lack editorial oversight.
“Many of these titles are cheaply produced mass-market items lacking editorial care or depth of content. These titles appear in the platforms’ search results right alongside original books produced by publishers with care and a commitment to quality,” he said.
“The sheer volume of AI-generated titles makes it harder than ever for readers to distinguish one from the other.”
Koch argued that without effective regulation, creators risk losing both control over their intellectual property and the financial incentives necessary to produce original work.
“Stronger regulatory frameworks are needed from the federal government and the EU to rein in Big Tech companies,” he said.
