Is It Just to Scan and Destroy a Book?

The book is one of the greatest tools that has passed down to generations the knowledge acquired by humans over thousands of years. It is a physical bridge that carries the thoughts of one generation to another, the experiences of one civilization to another, and the research of one scientist to future researchers. News that such books are being scanned for training artificial intelligence and then physically destroyed is now causing deep concern.

This is not a small matter of merely discarding old books. It raises the most serious questions: who should preserve the knowledge created by humans, who should control it, and in what form should it be passed on to future generations?

What is happening in America?

Recently, booksellers in America report a huge increase in purchases of old and rare books. Investigations have revealed that unknown buyers, purchasing large numbers of books at once, are scanning them at high speed, cutting off their spines, and ultimately destroying the physical copies. In some cases, reports say thousands of books are being bought at a time.

Is this true?

Yes, this is by no means a hypothetical matter. According to details that emerged in legal documents related to an AI company called Anthropic, a project named "Project Panama" involved purchasing millions of books, removing their bindings, scanning them, and subsequently "disposing of the physical copies." Reports also mention that one internal plan aimed to "destructively scan" all the books in the world.

But who is responsible for this crime?

Here, I want to make one thing clear. It is not correct to say that "AI has the intention to destroy books." AI does not independently have such desires or intentions. It is humans who make such ignorant decisions. In this matter, it is not appropriate to blame AI. AI companies are choosing such methods for training their models because they need vast amounts of high-quality, human-written data.

However, the question does not end there. Needing data for AI may be a necessity. But destroying the physical evidence of human knowledge for that need is morally unjustifiable, no matter how you look at it.

This should never be done, generally speaking:

When a book is scanned, its letters, words, sentences, and chapters are converted into digital form. But a book is not just its textual information. Its paper, printing method, typeface, cover, binding, publisher details, author's signature, readers' handwritten notes, old stamps, dedications, and the marks of time on its pages—all of these are information for future historians, librarians, and researchers.

A digital file can preserve a book's "content," but the actual book also preserves its "history."

That is why I am not against scanning. In fact, digitization is a great revolution in the preservation of human knowledge. It is an unprecedented advantage that a researcher in any corner of the world can read a manuscript thousands of kilometers away on a computer or mobile phone. UNESCO also uses digitization as a key tool to preserve documentary heritage and provide worldwide access to it, digitizing millions of pages in its archives and making them available to the public.

But we must not forget another important point UNESCO makes. Digital heritage is also not permanent. Digital files, storage devices, software, and hardware—all change over time. To pass digital materials to future generations requires constant maintenance, migration, and security.

This means it is not a simple matter to burn or shred a book and say, "Its information is on the server, isn't it?"

Because the server is also a system created by humans. It is not eternal. A company can shut down. A data center can be damaged. A file format may become unreadable in the future. Ownership may change. A company may restrict public access to digital information. Then, a dangerous situation arises: the knowledge created by humanity may disappear from the physical world and remain only on the servers of a few private companies.

This is a deeply concerning issue for humanity:

Knowledge has always developed as the common heritage of humanity. One scientist built upon another's research. One writer drew inspiration from previous writers. One generation's experiences reached another through books. This continuous chain of knowledge is the foundation of human civilization's progress.

Now, as this same knowledge becomes training data for AI models, the companies collecting it must answer a question: are you using the knowledge created by humanity, or are you gradually turning it into your own private intellectual property? And for that purpose, are you altering or destroying the original form of the book or document?

This is not an anti-AI question. This is a question about the direction in which AI development should proceed.

If AI is truly a technology beneficial to humanity, it should make human knowledge even more widely accessible. A book should not gather dust in a library in one country; it should be available to every student in the world. Rare research should be usable not just by one university but by researchers globally. This is the great humanitarian benefit that AI can achieve.

But why is it necessary to destroy the book to achieve this?

There are many methods for scanning. Technologies exist that can scan books without cutting or damaging them. Organizations like UNESCO follow the approach of expanding access through digitization while preserving physical heritage.

However, from a business perspective, the story looks different. Cutting a book apart and scanning it at high speed may be faster and cheaper. For AI companies needing billions of pages of data, carefully scanning each book while preserving it might be a costly process. But being cheap does not prove that an action is morally right.

Another question also arises here. The company that buys these books may legally have the right to destroy that physical copy. But is everything legal always moral? We should not conflate these two questions.

An individual can destroy their own book. But rare books, historical documents, manuscripts that survive as a single copy, and objects of cultural heritage possess a collective value beyond the limit of personal property.

Here, governments, national libraries, universities, archives, publishers, and AI companies need to come together and establish an international ethical standard.

"Digitize—but do not destroy rare physical knowledge."

That should be the minimum standard.

There is another, more important fear we have. Tomorrow, if our children and grandchildren ask, "Where are the books of the twentieth century?" we should not be in a position to say, "We scanned them all and put them on AI servers." Because we do not know—nor does anyone—whose servers they are, who controls them, or what rights the public will have over them in the future.

"Humans should not simply hand over their knowledge to machines. Machines should be servants to knowledge; they should not become masters of it."

That is why, in my view, we do not need to stand against AI. We should instead advocate for using AI in a more humane, transparent, and democratic manner.

Converting a book into digital form is giving a second life to human knowledge. But destroying the original book after scanning—especially rare, historical texts, irreplaceable manuscripts, diaries, palm-leaf manuscripts, and rare books—could cause irreversible damage to human knowledge for generations to come.

In our view, there is only one correct principle: "Scan and preserve; do not scan and destroy."

AI should not become a technological monster that devours human knowledge, but rather a technological tool that shares human knowledge more widely. Because books are not just paper. They are the memory of humanity. They are evidence of human thought. They are physical remnants of our civilization.

AI should build the future. But in the process of building that future, it should not have the right to destroy humanity's past.

Finally, here is the question I wish to ask:

Using the knowledge that humans have painstakingly created over thousands of years for the development of AI may be progress. But is destroying the physical embodiment of that same knowledge—the books—also progress?

It is not AI that must answer this question—it is the humans who are thoughtlessly destroying them, or doing so for the sake of conspiratorial business profits.