AI-Generated Books Flood Belgian Shops Without Labels

Thousands of books entirely generated by artificial intelligence are being sold on popular Belgian and Dutch webshops—including Bol.com, Standaard Boekhandel, Amazon, Bruna, and De Slegte—without clear warning labels indicating their AI origin, according to an investigation by the Dutch newspaper Trouw. Consumers are largely unaware they are purchasing AI-generated content rather than works by human authors, prompting alarm from authors’ organizations and librarians across Belgium and the Netherlands.

The Scale of the Problem

The company behind the flood of AI-generated titles is Andries B.V., a Dutch publishing firm based in Hendrik-Ido-Ambacht, Zuid-Holland. Since 2025, it has published over 2,000 non-fiction titles—producing approximately 10 new books per day. Topics range from “Alles over Straatsburg” (Everything About Strasbourg) to “Verticale tuinen voor beginners” (Vertical Gardens for Beginners), all generated entirely by AI with no meaningful human editorial oversight.

The books are produced using a print-on-demand model, meaning they are only printed when a customer places an order. This eliminates inventory risk and allows the company to offer thousands of titles without traditional publishing gatekeeping. Most books sell for approximately €16.95 each.

No Warning for Consumers

Until the Trouw investigation broke the story, most webshops did not include any visible indication that the books were AI-generated. Only Bol.com had a mention, hidden under “other characteristics” in product specifications. After the investigation, Bruna and De Slegte added descriptions noting the books were “created with the help of AI.”

Giovanni Coudeville, Head of Product at Standaard Boekhandel, told VRT NWS that his company prefers books written by real authors but is not in a position to determine what constitutes real literature. “That AI use must be clearly indicated,” he said. “That’s how you’re transparent to the reader.” Coudeville noted that publishers and authors should self-declare AI use in a central data platform, but currently, no such system is in place.

Authors’ Union Sounds the Alarm

Noor van der Heijden, spokesperson for the Dutch Authors’ Union (Auteursbond), described the situation as “unprecedented” and “alarming.” In comments reported by HLN, she warned that AI models are trained on copyrighted material without permission. “Those texts were stolen from our writers, and now such an AI bookmaker is competing with the writers it stole from,” she said.

Libraries Struggle with AI Influx

Flemish librarians first raised the alarm in March 2026 when hundreds of AI-generated books appeared in public library collections. Elisabeth Francet, a librarian in Landegem, Belgium, warned colleagues on social media about the influx. According to NOS, at least one Andries B.V. title was found in a Dutch public library in Rotterdam, where it was subsequently removed after staff added a note identifying it as AI-generated.

NBD Biblion, which helps libraries select books, has stated that AI-generated books often fail to meet quality standards and are being rejected. However, the sheer volume makes manual screening difficult. Of the approximately 25,000 Dutch-language books published annually, libraries select about 15,000—and screening each for AI generation is a growing challenge.

EU AI Act Looms

The revelations come just months before the European Union’s AI Act transparency rules are scheduled to take effect in August 2026. Under Article 50 of the AI Act, providers of generative AI must ensure that AI-generated content is identifiable, and certain content—including text published to inform the public on matters of public interest—must be clearly and visibly labeled.

Paul Hermans, Director of Literatuur Vlaanderen (Literature Flanders), told VRT NWS that a mandatory label is “absolutely necessary but doesn’t exist today.” His organization, which coordinates the BoekenOverleg (Books Council), is advocating for three key measures: a mandatory AI label on AI-generated books, a European levy on the use of copyrighted texts by AI companies, and a ban on granting copyright to texts wholly or partially generated by AI.

“AI is part of the future, we’re not naive about that,” Hermans said. “But only if that future is built on transparency, fair compensation, and respect for human creativity.”

Broader Implications

This case highlights a growing regulatory gap in the publishing industry. Andries Herremans, the owner of Andries B.V., acknowledged to Trouw that his books are made by AI, claiming a team “monitors closely” and steers the process—though he admitted no human expert reviews each title for content quality.

The story is part of a larger trend of AI-generated content flooding publishing markets. Similar concerns have emerged in academic publishing, journalism, and educational materials. As the EU AI Act’s implementation approaches, the question remains whether its transparency rules will be sufficient to address the labeling gap—and how many other “schijnuitgeverijen” (fake publishers) are currently operating under the radar.

What’s Next

With the EU AI Act’s transparency obligations set to take effect in August 2026, pressure is mounting on both platforms and publishers to implement proactive AI detection and labeling systems. For now, consumers shopping for books on Belgian and Dutch webshops are advised to check product descriptions carefully—because what looks like a human-authored book may have been written by a machine.