Were Your Books Stolen in Meta’s "Massive AI Training Book Heist"?

So instead of asking authors and paying for the rights to use our books to train their AI system, Meta evidently just used books and articles held in Library Genesis -- a "pirate" site.

Here's an exercise, courtesy of Alex Reisner at The Atlantic: Search LibGen, the Pirated-Books Database That Meta used to Train AI

When I ran the search on my own name, three of my published books came up (one of them twice.)

Run your own name, and see if your works -- without permission, without compensation -- were used to train Meta's AI.

The Authors Guild is calling it a "massive heist," and explains some of the things authors can do, including the information that "Class Action Lawsuits Cover All Writers Whose Books Were Used." They also say, in Meta’s Massive AI Training Book Heist: What Authors Need to Know:

"Meta and other AI companies knew exactly what they were doing but they did it anyway. Why? Because they needed books for their quality writing, style, expression, and long-form narration and would rather steal them than ask and pay for them as they do for all of the other necessary components of their AI, such as electricity and programming."

When I've used words like "Stolen" to describe the action of taking something without permission and without compensation and using it to create something else that is then sold, I've heard the response that I'm exaggerating.

I've also heard some folks say that the AI companies couldn't possibly afford to do it any other way. Except, for 2024 Meta reported, as Yahoo Finance explained, revenue of $164.5 billion -- and a profit of $62.4 billion. 

Billion.

And there are companies who are licensing works for AI training (in fact, the Authors Guild is working with one.) Note: Consent of the creators is key.

Bydonmartin at Instagram (ironically enough, as it is a Meta-owned platform) summed up their creator perspective here. It's well-worth watching.

It's all pretty upsetting, but helpful to stay informed. And maybe, as the Authors Guild suggests in their list of five things authors can do right now, taking some action can help. Here's their first suggestion:

Send a formal notice: If your books are in the LibGen dataset, send a letter to Meta and other AI companies stating they do not have the right to use your books. Here is a template you can use. 

Writing the letter took me two minutes (just added my info and my stolen book info) -- I've copied the text of my letter below. Did it make me feel a bit better? Yeah, a little. And if enough of us do it, maybe it will have an impact. You can draft your own letter here.

Illustrate, Translate, and Write On,
Lee

--

To: Meta CEO Mark Zuckerberg

Subject: You Do Not Have the Right to Use My Work

My Message:

I am writing to you as an author who is extremely concerned about your misuse of my creative work. It has come to my attention that you have used my books, Queer as a Five-Dollar Bill, A Different Kind of Brave, and No Way, They Were Gay? Hidden Lives and Secret Loves, in the training of your generative AI models without permission from me, and in violation of my rights under copyright.

This letter is to put you on notice that you do not have the right to use my work to train your AI models. You must obtain express permission and provide reasonable licensing terms for authors’ works.

I hope you will set an example of responsible, legal, ethical AI use by obtaining permission before using authors’ and journalists’ works going forward and compensating us for the use you have already made. 

Lee Wind