Anthropic Thievery

A searchable database of one of the pirate websites Anthropic used to train its product lists 67 works under my name.

67.

Books, short stories, some nonfiction…there’s a fair bit of duplication, but that’s a lot of my words and sentences stolen to train these things. (Apart from the sort of helpless anger one feels at realizing everything gets pirated to begin with.)

The Anthropic settlement is a sort of good start, I guess, but that amount of money is just a cost of doing business for them, and the settlement imposes no meaningful safeguards on other companies building their products on the theft of actual humans’ actual work.

One response to “Anthropic Thievery”

  1. We found 8 of our books in the LibGen search tool, only two of them were in the date range referenced in the lawsuit though. I’ve had a number of exchanges with authors and artists about this suit in the past month or so and something that keeps coming up is the vast difference in “$3000 per work” and the “$60000 per song” that regular, broke teenagers go whacked with in the Metallica piracy suit from years ago. Referencing that to Sam Altman talking about burning $115 BILLION in the next 3-4 years it’s a slap on the wrist indeed to companies of this size. The cherry on top is that this suit doesn’t even set a precedent since they rolled over so fast with a settlement.

    I guess we can at least find some solace that this case was won, even if only in settlement, and that it’s only for works pirated between 2021 and 2022, that had an application submitted for copyright registration filed with the US Copyright Office before or during the scope of those years and within five years of publication of the works. So hopefully we’ll see some more suits with broader scopes or many different suits with many different scopes that hopefully also go to judgement to set a precedent. Tempered with the current administration in mind though of course.

    Best Regards, Chris

    Like

Leave a comment