Sony Music Publishing and Warner Chappell filed suit against Anthropic on 28 August. Thirty-five publishers joined it. The complaint runs 48 pages and names the company, its chief executive Dario Amodei, and its co-founder Benjamin Mann. It says they obtained songs by torrenting and scraping, and used them to train Claude. Anthropic says the case recycles allegations already before the courts and that it will defend itself.
Five weeks earlier, on 20 July, a court gave final approval to a different settlement. Anthropic agreed to pay 1.5 billion dollars to a group of book authors. The order covered 482,460 works. That comes to roughly 3,000 dollars a book.
I have written seven books. One of them is published. It is called Unfold and it sits on Amazon. I will not see any of that money. Neither will most writers reading this.
The settlement paid for the theft, not the training
This is the part that gets lost in the coverage.
The court in that case found that training a model on books was fair use. What survived was how the books were obtained. Anthropic had downloaded millions of them from Library Genesis and the Pirate Library Mirror. The settlement was about acquisition.
So the pool of authors who got paid was defined by piracy. Not by authorship, not by sales, not by quality. If your book had been scanned, uploaded and indexed on a pirate library, you were inside the class. If it had not, you were outside it.
Pirate libraries mirror what moves through the trade. Print runs, publisher catalogues, files that circulate because a distributor put them somewhere first. A book that exists only as a listing in a self-publishing dashboard rarely gets there. It was never important enough to steal.
That is the position I am in, and it is the position most self-published authors are in. We were not worth pirating, so we are not owed anything, so we are not in the room where the money is being divided.
Where I actually stand
I am not against a model reading my work. If the result were free to people and no company earned from it, I would have no complaint at all.
That is not the situation. The companies charge for the output. The material that made the output possible was taken without asking and, in the cases now in court, taken from pirate sites.
I also think the people demanding payment are right, and I say that as someone who would not qualify for a cheque. It was their work. They made it. Someone else built a business on top of it and did not ask.
There is a test I keep coming back to. Anthropic would not accept the same treatment. If someone torrented its model weights, took its source code, trained on them and sold the result, that company would be in court within the week. The argument about learning and transformation only ever seems to run in one direction.
I have used these tools. I used one to score my own manuscripts against criteria I set, and the rankings came back reasonable. So I am not writing this from outside the thing I am describing. I am inside it, using it, and still able to see what was done to get here.
What a self-published author can actually do
Not much, and I would rather say that plainly than sell hope.
Check whether you are in the class. The Bartz settlement had a claims process with a list of covered works. Search for the settlement administrator’s site and look your titles up. It takes ten minutes and it removes the doubt. Most people reading this will not find themselves there. A few will, and they should file.
Register your copyright if you are in the United States or publish there. Registration is what makes statutory damages available. Without it you are relying on someone else’s class action to sweep you up, and these classes are drawn from pirate library indexes, not from author lists.
Stop treating the lawsuits as your route to being paid. They are the trade fighting over the trade. The publishers suing are the same organisations that turned down seven of my manuscripts over four years without ever putting a book of mine into the world. Their interest and mine are not the same interest.
Then go back to distribution, because that is the only part of this you control. Being read is still worth more than being compensated for having been copied.
The uncomfortable version
There is a reading of all this that stings, and I think it is the true one.
The value of a book, in these settlements, was set by whether someone bothered to pirate it. Roughly 3,000 dollars if they did. Nothing if they did not.
By that measure my seven manuscripts are worth nothing at all, and the six that are not published are worth less than nothing, because they do not exist as far as any of this is concerned.
I do not think that is what a book is worth. I do think it is what the market has just told us, in writing, with a judge’s signature on it.
