- Trying!
- Posts
- Why did you steal my book, Mark Zuckerberg?
Why did you steal my book, Mark Zuckerberg?
Come on, it was 50% off on Amazon! You didn't have $10.50?
Matt Gross
March 26, 2025
In partnership with
Today’s advertiser is, once again, Authory, whose automated portfolio system I’ve subscribed to for years now. Although Beehiiv rules prevent me from asking or encouraging you to click on Authory’s ad, if you do click on it down below, of your own free will and according to your own moral principles, each click will earn me $2.
My travel memoir, The Turk Who Loved Apples, has in the nearly 12 years since it was published, sold 625 copies. That’s it: 625. According to my most recent royalty statement, 332 of them were paperback, and 293 were electronic. This doesn’t include the Chinese translation of the book, which came out in late 2018 or early 2019, nor the handful of copies I sold at the Read the Room literary salon, so it’s entirely possible that a whole 1,000 people bought The Turk Who Loved Apples.
One person who did not buy it: Mark Zuckerberg, who I’m told is the CEO of Meta, owner of Facebook, Instagram, WhatsApp.
He did, however, acquire it. According to a report last week in The Atlantic (which is just killing it lately, right?), his minions, in their rush to train a new AI model, Llama 3, on “high-quality” writing, decided that legally acquiring books—such as the reasonably high-quality one I wrote—would just take too long:
Meta employees spoke with multiple companies about licensing books and research papers, but they weren’t thrilled with their options. This “seems unreasonably expensive,” wrote one research scientist on an internal company chat, in reference to one potential deal, according to court records. A Llama-team senior manager added that this would also be an “incredibly slow” process: “They take like 4+ weeks to deliver data.” In a message found in another legal filing, a director of engineering noted another downside to this approach: “The problem is that people don’t realize that if we license one single book, we won’t be able to lean into fair use strategy,” a reference to a possible legal defense for using copyrighted books to train AI.
So instead of waiting “like 4+ weeks,” they decided to steal. They turned to LibGen, a library of pirated material:
It currently contains more than 7.5 million books and 81 million research papers. Eventually, the team at Meta got permission from “MZ”—an apparent reference to Meta CEO Mark Zuckerberg—to download and use the data set.
And that’s how we get this:
More after the highly clickable ad…
🪨
Writers, don’t let your work disappear!
Imagine losing years of articles because a site shut down. What would you do if all your work samples disappeared?
With Authory, that’s a nightmare you’ll never have to face. Authory automatically creates a portfolio that backs up everything you’ve ever written and will write, so your work is always safe.
That’s right: Authory finds and backs up all your past work and saves every new piece you publish, wherever they appear.
Join thousands of writers who already trust Authory to protect their work and never lose a piece again.
🪨
Now, I’m not necessarily a copyright absolutist when it comes to AI models. Don’t get me wrong—I loathe AI, and I look forward to the bursting of the current AI bubble, although it may wreck a huge chunk of the economy.
But is training an AI on a book—say, The Turk Who Loved Apples—an actual violation of copyright? Perhaps, if specific sentences and paragraphs wind up reproduced verbatim in that AI’s responses. But I’m not sure an AI model consuming all those books is massively different from me reading loads of books and movies and songs and then sprinkling references and allusions to them throughout my own writing—or even just being generally influenced by their language or atmosphere. Am I violating copyright by quoting Björk or apostrophizing Nabokov? Or is it a violation only if I stole the Björk CD?
I suppose there’s probably a real legal argument to make against anyone—Meta especially included—who has used an author’s work to train an AI model without the author’s explicit permission. Something about terms of service: that is, what rights do you actually purchase when you purchase a book? Do they include the right to train an AI model and/or your own literary sensibility? And so if you’ve stolen the book, then you haven’t properly acquired those rights, right?
But there are those who would argue, too, that even properly acquiring a book does not convey the right to use it to train an AI. Which I’m not too sure about—and which leads us into legal territory I am not qualified to analyze. At least, not yet!
All the legal stuff aside, this hurts because it’s insulting. Look, I know no one bought my book. It’s a little weird, and the sentimental title really doesn’t prepare a reader for the temporospatial and emotional hopscotching within. Most of the reviews were ho-hum. But I do think it’s fundamentally a good book, and I worked fairly hard—not really, really hard, but fairly hard—to make it a satisfying and unique read. And still, Meta and Mark Zuckerberg, with all of their hundreds of billions of dollars, could not wait a few weeks and spend $21.99 (50% off on Amazon right now!) to get their grubby digits on The Turk Who Loved Apples?
Meta is, of course, facing lawsuits galore over this, and you’ll surely be hearing about the Stephen Kings and J.K. Rowlings whose books were pirated on LibGen and used by Llama. But I’m more concerned with authors like me, those in the Under 1,000 Club, who have struggled to get even a bit of notice for our work—and who then watch enormous corporations hoover it up without even the pretense of “licensing” it for consumption.
Dick move, Zuck.
In other words, I’m very happy to join any lawsuit that will make them pay for this abuse. After all, it may be the only way The Turk Who Loved Apples ever earns a cent.
Meanwhile, the print edition of Trying! has, as far as I can tell, not yet been pirated by LibGen. (As usual, I have mixed feelings about this: What, am I not good enough to be stolen?!?) You are, of course, welcome to acquire your own copy for a mere $12.34—it is fun and easy to read, but the single best thing about the book, apart from the genius illustrations by Aliza Gans, is that it fits nicely and neatly in almost any pocket. Whether you use it to train an AI or just your own literary philosophical sensibility is up to you. But maybe first consult with your lawyers. 🪨🪨🪨
Reply