Big Tech quietly buying up Dutch secondhand books for AI training

Over the past few months, several Dutch secondhand book sellers have noticed something unusual: large, coordinated orders for Dutch-language books arriving from abroad. Experts believe these books are being collected by major artificial intelligence (AI) companies — the firms behind chatbots and language systems like ChatGPT.

Here's what's happening: AI companies train their systems by feeding them vast amounts of text — books, articles, websites — to teach the software to recognize patterns and generate human-like responses. The more diverse and high-quality the text, the better the AI performs. Dutch-language books, especially out-of-print titles, are apparently valuable for this training because they provide authentic Dutch writing that's not easily available online.

The concern is significant. These book purchases appear to be funneled through intermediaries or bulk buyers, meaning the original authors and publishers rarely — if ever — know their work is being used this way. They certainly aren't being paid for it. One observer called this "a form of barbarism," arguing that it's essentially theft of intellectual property on a massive scale.

For everyday readers and authors, this raises a real question: should creators be compensated when their work trains the AI systems that now influence so much of what we see online? Right now, there's no clear answer, and no agreement on fair payment. It's a growing ethical problem that regulators and tech companies have barely begun to address.