Anthropic's $1.5B Settlement Comes With a Broken Claims Process
Anthropic owes roughly 500,000 authors $1.5 billion for downloading millions of pirated, copyrighted books to train its AI model without permission. But the website authors must use to claim their share is so buggy that many can't navigate it.
The settlement, handed down last fall, marked the first major legal reckoning over generative AI and LLM training practices. A judge found that Anthropic's use of pirated books without authorization constituted copyright infringement, though the company argued the training itself qualified as fair use. The settlement requires Anthropic to pay authors for the unauthorized use of their work.
The Claims Problem
The problem is execution. Authors like Maureen Johnson, who has published 18 books including several bestsellers, found themselves wrestling with Anthropic's claims website. Johnson stands to receive roughly $3,000 per book affected by the settlement, to be split 50-50 with her publisher.
The data set Anthropic built to process claims is so error-ridden that authors struggle to confirm their books were included or calculate what they're owed. The website itself functions poorly, leaving writers unable to complete the claims process.
One author's frustration captured the broader sentiment: "Your AI monster ate all our work. Now you're trying to pay us off with this piece of garbage that doesn't work."
What's Next
Similar copyright suits are pending against Meta and OpenAI. Both companies face the same core allegation: they trained AI models on copyrighted material without authorization.
For writers affected by these settlements, the legal victory means little if the claims process itself becomes another barrier to compensation.
Your membership also unlocks:
