AI stole my book and sold it online

Someone used AI to copy my first book and is selling it on Amazon

If you look forward to reading this newsletter. Become a paying subscriber. Paying subscribers are the reason this newsletter exists. So thank you.

If you search for my book “This American Ex-Wife” on Amazon you will find not only my book but a series of “This American Ex-Wife” workbooks all marketed as “detailed and practical” guides to my work. 

These are AI-generated summaries, distilling down the essence of my 249-page book into bullet points and paragraphs, flattening nuance and removing quotes and citations.

Summaries and study guides of popular books aren’t a new phenomenon. After all, whomst among us hasn't relied on Cliff’s Notes while reading Uncle Tom’s Cabin in high school English? But the problem is different. Before, a Cliff’s Notes summary would be reserved for works considered classics. Now, AI-generated guides exist for any and all books that the algorithm deems worthy. And here, worthy means, they can make a few bucks off it rather than any sort of intrinsic literary value.

It’s a destabilizing kind of revolution. The white male academic colonial ideals are swapped out with crass consumerism. Money, notoriously, isn’t a great equalizer, so things aren’t getting better — just more chaotic.

Once upon a time, Cliff’s Notes were written by poorly paid humans. I’m not saying that was better. Humans, especially when poorly compensated, introduce errors and flatten nuance, just as AI does. But the lack of cohesion and the copy that reads as if a 3rd-grader had learned to use a thesaurus but not well makes the AI style of rendering, its own kind of unnatural prose that parses only on a cursory glance. But presumably a cursory glance is all that copy is getting.

Other authors have tried to fight to take down these summaries before. It ends up being a game of internet whack-a-mole. You take one down, another pops up. Also, workbooks like this haven’t lifted enough word-for-word material from the book to constitute a copyright violation; if they have, who is going to buy the workbooks and scan them line by line? A lot of these books aren’t available for Kindle or e-reader. So you’d have to buy the print-on-demand paperback and either scan the copy and compare it to yours or read it yourself. Who has the time?

But as I scrolled through this algorithmic chum bucket, I saw a biography titled Lyz Lenz: A Story of Resilience and Reinvention. The cover is eerily similar to the cover of my first book God Land. And I wondered what it contained. Was it an AI-generated biography? Did it smush together the writings of my past three books? Scrape information off the internet?

The Amazon summary described me as Catholic. And I have never been Catholic, although I did spend a lot of my elementary school years pretending to be a nun who got stigmata. This book, too, was not available for e-reader, so I paid the $14.50 to get a copy. It came two days later, a cheaply printed 138-page book that was almost exactly a word-for-word reprint of God Land.

The book had been run through the word chipper of the AI machines — rasterizing and re-rendering my words into some sort of uncanny valley version of the original. Where I wrote “city,” the book had “metropolis”; where I wrote “said,” the “biography” had “murmured.” Proper nouns that should be capitalized were lowercase, and most of the longer quotes and secondary research had been stripped from the book. 

There were no footnotes, no acknowledgements — although, hilariously, there is a copyright page insisting that “the contents of this book may not be copied, reproduced, or transmitted without the express written permission of the author or publisher.” 

The author of this book, Deborah A. Otten, seems to be a fake name. My search of Deborah A. Ottens only turned up a handful of young grandmothers who seem to be more likely to sell Mary Kay than be involved in an AI copyright violation scheme. 

In 2023, The Atlantic published a list of books that had been used to train generative AI. God Land was one of them. In response to the investigation, Sarah Silverman and other authors filed a lawsuit against Meta, alleging that the use of their books constituted a copyright violation. That case is still working its way through the courts and in February it hit a stumbling block when a judge dismissed one of the claims that essentially argued that since ChatGPT was trained on copyright material, all the content of ChatGPT produced was an infringement.

This lawsuit is similar to another one filed by a group of novelists including George R. R. Martin and Jodi Picoult. In this lawsuit, the authors specifically mention AI-generated rewrites of original works that, much like the biography of me, borrow heavily from copyrighted material.

These lawsuits will take years to work through the legal system. In the meantime, AI content machines will continue to crank out the cheap knock-off versions of better books, disguising them as workbooks and biographies. The result is that Amazon is now a seedy street vendor of both real and black market goods.

What does this mean for the future of writing? I spent part of this weekend skimming through the biography of me. Wondering if I, the person who’d written the book, could spot some of the differences or if some of the changes had even been good. It felt like looking at a clone of myself — one with dead eyes or hands that move in an unnatural way. I had a sense of unease. I could spot some of the differences, but not all. I didn’t think it read better than the original, but was I even the best judge? It was me but almost. My words but not quite. A gross kind of verisimilitude.

Mostly, I was just bored. Whatever fun and life had been on the pages of the first book seemed leached from the pages of this poorly produced knock-off. 

But it was worth wondering whether this writing was better. If falsity could be its own kind of truthlikeness. If, as Karl Popper asserted, that falsity could sometimes be closer to the truth.

If I were George R. R. Martin and had a swimming pool of money, I would probably first buy an actual swimming pool. But second, I might hire someone to scan the copy, do a side-by-side comparison, hire a lawyer, and issue a takedown notice. But I do not. And it’s summer and I don’t have childcare. I emailed the publisher of the first book to let them know and they said they’d look into it — but confessed that it’s hard to send a cease-and-desist to a fake author, fronted by what is probably a fake company, selling print-on-demand, on a massive website that has virtually little oversight or accountability.

I’m not exactly outraged by what appears to be a copyright violation. I am not clinging to my words at night sobbing, “My precious sentences! How dare they!” (I am not above that kind of behavior, and have in the past gone to the mattresses with editors over wording choices.) 

But I am not angry about this. It seems almost inevitable. 

Given that our timeline is a shitty rewrite of the past with only a few minor words changed, a shitty regurgitation of one of my books is just par for the course.

My dad is a copyright attorney who for many years worked for companies like Texas Instruments and Gateway Computers. If you put his name into Amazon, the results will return a copy of his book Representing High-Tech Companies, which sells for $906. There are no AI knock-offs. I don’t think the algorithm is even interested in reading it. I doubt the few lawyers and law students forced to slog through the 1,500 pages are interested in reading it. At home, we joke it’s less of a book and more of a doorstop. And while my parents bought 30 copies of my most recent book, I have never read my dad’s.

I don’t often call my dad for legal help. Once I had him make sure I was interpreting pages and pages of Miss America lawsuits correctly. (I was.) And he did help me form my S Corp. The few times I’ve needed representation, I’ve hired it. But this is exactly his niche. For years, he’s worked protecting copyrights and defending against violations. 

So, as I racked my brain for someone to talk to, I figured this was an in-house job.

Carl Baranowski compares this new world of knock-off books to knock-off Prada. The Prado you could buy on the streets of a big city. The original but now only cheaper and more accessible. These AI-generated books are the Prado of books. Books are now schmooks. Lyz Lenz is now Ryz Renz.

He tells me that in some way this is a tale as old as time, and reminding me that the iconic Obama “Hope” poster was a rip-off of an AP photo. People are always taking something and turning it into something else. But the legal standard is: Is it transformed? 

AI-generated content can’t be considered under copyright, but there are exceptions when the author has worked to transform the AI content.

It’s a wild world right now. Like the one we were in when Napster and Limewire made their appearance. Laws will catch up, the marketplace will shift, but AI will never go away. I tell my dad about how AI is turning my profession to crap with the proliferation of shitty content across media platforms. But then I recall that in my early 20s I wrote that shitty content — cranking out unbylined blog posts for content machines, just as a way to earn money.

Recently, Amazon dialed back a program that was supposed to be AI-generated, but was actually run by humans. 

My dad thinks AI is a good thing. I ask him to name some examples, and the best one he comes up with is Grammarly¹. He also enjoys the AI tool that allows him to search WestLaw.

I am, in some ways, my father’s daughter. I am not convinced that AI is the downfall of humanity exactly. In fact, sometimes, I very much would like to have some humans replaced by robots. I think self-driving cars will be great one day. I love to order McDonalds from an app or at the large-screen kiosk. 

But I am concerned about a world in which everything, every word and idea, gets thrown into an AI-generated Ninja blender until what comes out is some tasteless beige baby-food word puke that has no relation to ideas, reality, or even truth. 

We live in a world where nothing is working. Our democracy. Our courts. Our highest levels of government, yes. But even basic functional elements that we could count on before are falling apart. 

Websites like Twitter take forever to load. Pop-up ads clog the screen. Automatic video screeches out from the speakers. Ads for stories about people with three nipples fill the copy of stories I am trying to read. Calling customer service is a nightmare of automation. My new washing machine started leaking recently and it took three months of filling out online forms and calling before I could even speak to a human about fixing it. And even then, the human had me repeat what I’d told the bots before him. I wasn’t very nice about it. 

We poor humans are so exhausted by the inefficiencies of automatization that when we finally reach one another we can barely speak civilly. Everything has been optimized for commerce and nothing is optimal. Nothing even really works.

Other thoughts:

• This is a beautiful essay written by Vauhini Vara and an AI bot about the death of her sister.

• I thought this essay by Jay Kaspian Kang about using AI to write his novel was interesting.

• Me and Savannah Guthrie having the same problem.

1. Grammarly is amazing, but my human editor is even better.