You’ve probably been told that AI is just “reading” the internet. That it’s digesting text the way a college student cramming for an exam does. But what happens when the AI doesn’t just read the book—what if it has to destroy it to learn from it?
Recently, 404 Media tracked a shipment of rare books. Where did they end up? An Amazon AI training facility. These aren’t being archived. They are being ripped apart, scanned, and discarded to feed the data extraction machine. And the most chilling part? The law completely encourages it.
We built a legal system that protects the copy, but leaves the original to rot.
Look at the Internet Archive. They tried to lend out scanned digital copies of books they physically owned, acting exactly like a traditional library. The publishing industry hit them with ruinous litigation, dragging them through the courts and calling it piracy. A nonprofit trying to keep culture accessible is treated as a threat to society.
In the eyes of copyright law, a trillion-dollar corporation destroying a physical artifact is “fair use,” but a nonprofit library trying to keep it accessible is “piracy.”
Here is the twist most people miss: copyright law is built around protecting reproductions, not physical artifacts. If you make a copy, you’re a pirate. But if you buy the original, extract its data, and throw the carcass in the incinerator, you’re just a consumer exercising your property rights. The destruction of the book isn’t an accident. It’s the logical endpoint of treating books as raw data.
The destruction of the physical book isn’t collateral damage—it is the logical endpoint of treating human culture as raw data.
Once the content is extracted, the physical copy has no legal status. Destroying it becomes legally invisible. We are watching powerful AI operations erase physical culture while the law shrugs, and we are letting it happen because we assume “fair use” actually means fairness.
This isn’t just about old paper. This sets the precedent for who gets to own and consume cultural heritage. If AI training can consume rare books under fair use, libraries lose the legal ground to preserve them, and the public loses access to its own history. We are trading the physical soul of our past for a slightly better chatbot. And the law is cheering the whole way.
FAQ
Q: Doesn't Amazon own the books? Can't they do what they want with their property?
A: Yes, under current property law, they can. But that's exactly the problem. The law grants absolute destruction rights over physical artifacts while criminalizing digital preservation. It rewards extraction and punishes keeping history alive.
Q: What does this mean for the future of libraries?
A: It means they are being outpaced and legally outmaneuvered by AI companies. If extracting data and destroying the original is legally protected 'fair use,' libraries have no ground left to stand on. The public loses access to history unless a corporation decides to monetize it.
Q: Isn't the information more important than the physical paper?
A: No, because the physical artifact is the only proof of historical context. Once you destroy the original, you centralize access to the data. You hand the monopoly of human history over to whoever owns the server.