The bookseller using AI to battle back against Big Tech’s digitize-and-destroy campaign 

HOUSTON — Becker’s Books smells exactly the way it’s supposed to — like dust so deeply absorbed into millions of paper pages that it has lost its nose-tickling edge, warmed by a touch of vanilla. There is a technical term for this scent: bibliosmia. But second-generation bookstore owner Charlie Becker describes it simply: “old book smell.” 

It’s the scent of his childhood. Becker “literally grew up here,” he says on a Thursday afternoon in early August as he wanders through the labyrinthine passages between sky-high shelves. 

He loves this smell. And he’s hell-bent on saving it for future generations. 

In 2026, that quest is becoming particularly challenging. 

A 70-book sweep

In April, Becker began noticing an uptick in his online sales. In the course of one week, he logged orders for two books, then three. 

“And then even four, five, six, seven-book orders, which is really odd,” he says. “Especially when the books are different types of books.”

It’s not as if one buyer snapped up every edition the shop has on one topic, like quilting or gardening. The orders spanned subjects.

READ MORE: Behind the local bookstore boom across the US

“These books are basically the type of books where it would be rare for someone to buy those online,” says Becker. “They’re more like the books people might buy in-person, because they’re books that you would not ever search for online.”

Think: A 1989 travel guide to Phoenix, Arizona. 

Becker flagged his parents, Dan and Ann, who first opened the store in 1993 and still run much of its day-to-day operations. Together, the trio decided to keep an eye on the situation before jumping to conclusions — including a growing conclusion among booksellers across the globe that AI companies are mass-purchasing old books to scan their contents and destroy their physical shells. 

After talking with his parents, Becker went to bed. 

“That night, we got a single order for 70 books, which is unheard of,” he says. 

Becker speaks in precision. He is careful to note that he can’t definitively say that one 70-book order has literally never been placed in the 32 years since his mother first began selling inventory online in 1994. “But,” he says, “we couldn’t remember a time.”

And Becker’s Books has a long memory. 

A bookselling blindspot

Housed in a barn-red 1950s ranch home, the 3,800-square-foot shop holds somewhere between 200,000 and 300,000 volumes — some tucked, others teetering — on its shelves. Six minutes away, an overflow warehouse doubles the inventory.

Many — most, perhaps — of the books were published before 1970, which is a significant year in the publishing industry’s timeline. Books published post-1970 are all equipped with an International Standard Book Number, or ISBN, a unique code that acts like a digital fingerprint. If 10 shops list a 1985 edition of “Dune,” the ISBN connects those listings behind the scenes, making it clear how many copies exist. 

Before the ISBN, there was no simple way to take a census of books on the market. A 1968 copy of “Dune,” for instance, lives in a silo online, disconnected from its fellow copies. That makes it impossible for sellers to know whether their edition is the last of its kind or one of dozens of remaining copies. 

This is creating a growing problem for the used- and rare-book industry. Earlier this month, a 404 Media investigation found that tech giants are buying rare books, shipping them to warehouse facilities and slicing off their spines before scanning in the text and discarding the carcasses. The goal, according to 404: Train large language models with text that’s guaranteed to be written by actual humans, which protects LLMs from “model collapse.”

Destroying the tactile book is a key part of the process. Once a buyer purchases a physical book, destroying the paper copy while digitizing its text provides a loophole in copyright law’s “first-sale doctrine,” ensuring the book has not technically been duplicated. 

Some sales are just not worth it, says Becker, who likens this process to a modern-day burning of the Library of Alexandria.  

So he’s building a tool to help booksellers know what they’re up against. And yes, he’s using AI to do it. 

Fighting code with code

Before returning to Becker’s Books full time, Becker spent a decade as an entrepreneurship professor at the University of Houston, teaching would-be business owners how to navigate operational friction. Last year, he applied for a $100,000 fellowship to fund a project that would build an AI-powered, open-source cataloging tool for independent booksellers like him. His proposal was one of a dozen selected by O’Shaughnessy Ventures from 7,000 applicants.  (It was an immediate yes, according to O’Shaughnessy Ventures Chief of Staff Jean-Marc Daecius, whose initial reaction to Becker’s pitch to solve an old problem with new technologies was, “Oh, this is awesome.”)

Becker is incredibly familiar with what he’s up against. Back in 2023, he was quoted in the New York Times as an early adopter of ChatGPT, which he used in his academic work. Now, he spends much of his time scanning covers and title pages into his growing catalog in an effort to block Big Tech’s digitize-and-destroy campaign. 

READ MORE: Social media isn’t so social anymore. What will platforms do next to grab our attention? 

“Our data from that profile creation will actually be the canonical record for a lot of these books once uploaded to the database,” Becker says. “We’re actually creating data. We’re not finding data.”

In other words, many of these books are being scanned into a system for the first time. His goal, once he can pick up the pace, is to create a sweeping catalog of rare and obscure books published before 1970.

It won’t be easy. The folks at ISBNdb, which catalogs more than 30 million ISBN-era books, told Straight Arrow that “identifying and counting those editions is genuinely difficult and projects that take that on are doing the book world a service.”

Once Becker’s catalog is created, other booksellers will be able to upload their own scans, creating a global database able to track how many copies exist of all these pre-1970 books. 

“If this works, in three or four or five years, people can easily, rapidly generate listings and contribute them to a master database where we’re going to keep track of what’s where,” says Becker. 

Any seller who receives an order that raises questions can consult Becker’s database, “where they can at least be like, ‘Oh my God, am I the last person who has this book?’ before they sell it,” he says. “And they can make an informed choice about whether they want to sell it or not.” 

Changing the ‘how’

For the most part, Becker tries to steer away from philosophizing over who “gets” to buy a book. But he does set some ground rules on what makes a book a book.

“I like to think of books in three ways: One is a book as knowledge; then book as experience; and book as an artifact,” he says. 

Outside of those purposes, which encompass everything from study guide to book club fuel and shelf trophy, books become something else entirely. 

“Generally, anything where it looks like a book’s being commodified beyond what a book is for, we’re kind of like, ‘Hmm, you know,’” he says, placing a finger on his chin. These situations make him pause and consider: Are these sales he wants to take part in?

“We even have discussions about, like, we don’t really like people who buy books by the foot for a color,” he says. Becker’s Books sells to interior designers, of course, but usually they assemble collections based on topic or vibe. “If they’re like, ‘I need 50 orange books,’ I’m like, ‘Why not just use tiles or something? Fanta?’”

Bottom line, he says: “We don’t want to be doing outside procurement for this other random business.”

After the 70-book order, Becker pulled his books from the platform entirely. Now, he’s doubling down on creating his database to help equip himself and other sellers with key context that could help them decide whether future sales are worth passing up. 

Growing up inside a house of books has taught him that knowledge is powerful. And he’s willing to test those limits. 

A small bookshop owner in Houston, Texas, is unlikely to stop the forward march of time. Becker knows this. But while he accepts that he can’t change what is happening in the AI revolution, he is buoyed by the knowledge that his work can change how it happens. 

“The development’s happening, and the money’s going in,” he says. “And if there are problems that you want to solve, and AI may help, I feel an ethical responsibility to really call as much attention to that as possible.”

So Becker scans covers and title pages, building a system to help bookshops protect America’s Library of Alexandria. 

A way to preserve bibliosmia for another generation.

Round out your reading

• All your questions about napping, answered.
• Trump’s $5,000 promise echoes past payouts that never materialized.
• Inside the effort to make data centers pay their share of electricity costs.
• Why did the Feds seize the ’largest Martian meteorite on Earth’?
• Photos and video show exactly where and when Trump visited Ground Zero after 9/11.