Anthropic, Books, and the Ethics of AI Training

I’ve been following the recent reporting around Anthropic’s book-training practices, including the Washington Post’s reporting on destructive scanning and Reuters’ coverage of the copyright settlement.

As the founder of Page & Provenance and as a collector, this hits close to home. Books are not disposable. They are historical records and beautiful physical objects that people care about deeply.

I care about books enough that even a small scratch bothers me more than I can explain. That is exactly why this conversation matters to me. If books are going to help build AI systems, then the people building those systems should be honest and accountable about what is being traded away.

What the reporting says

Recent reporting and court discussion around Anthropic have raised questions about how books were used in AI training and how physical copies were handled in the process.

That distinction matters. The legal side is important, but so is the cultural one. Books are not just text containers. They are editions, artifacts, and records of ownership, authorship, and time.

For collectors, dealers, librarians, and writers, that physical object matters. The margin notes matter. The binding matters. The edition matters. The copy itself matters.

Why this feels personal

As a collector, I know how much these objects matter. Even a small scratch stays with you. That may sound sentimental to someone outside the book world, but collectors understand it immediately.

A book can carry emotional, historical, and financial value at the same time. It can be a first edition, a family heirloom, a signed copy, or simply the book that changed how someone sees the world. When you care about books, you care about more than their content.

That is why the idea of books being treated as disposable feels wrong.

The ethics question

I’m not interested in blame games. I’m interested in ethics.

If books are going to help build AI systems, then the people building those systems should be honest and accountable about what is being traded away.

That means asking real questions:

• What happens when physical books are treated as raw material instead of cultural objects?
• What responsibilities do AI companies have when they rely on books to train commercial systems?
• What gets lost when efficiency becomes the only value that matters?
• How do we preserve the record while still making room for innovation?

These are not abstract questions. They matter to the future of books.

What Page & Provenance believes

We’re building Page & Provenance around the opposite idea: transparency, documentation, and respect for the book itself.

That belief shapes everything we do. We want to help people understand what they own. We want to help collectors document their books properly. We want to create systems that preserve provenance instead of erasing it.

That is the foundation behind HonestBookAI, Page & Provenance Collections, and the broader platform. The point is not just to make books easier to evaluate or organize. The point is to make the history around them more visible and more durable.

Books are not disposable

This conversation also deserves a simple public message.

Books are not disposable.

That is the idea behind #ProtectThePage. It is short, clear, and easy to understand. More importantly, it says something true: books deserve care. They deserve context. They deserve to be treated like part of the human record, not just another resource to extract value from.

Closing

I know AI and books will continue to intersect in complicated ways. That is not going to stop.

But complexity is not an excuse for carelessness.

If books are going to help build AI systems, the people building those systems should be honest about what is being traded away. And those of us who love books should keep saying that the physical object still matters.

Books are not disposable.