Anthropic thought it had bought itself some peace with writers, $1.5 billion worth, to be exact.
But Judge William Alsup isn’t ready to hand out the “case closed” stamp just yet.
In fact, he rejected the record-breaking settlement this week, saying it was “nowhere close to complete” and might be shoved “down the throat of authors.”
The lawsuit, filed by roughly half a million writers, accuses Anthropic of using pirated versions of their books to train its large language models.
The proposed settlement promised $3,000 per work, a payout one lawyer gleefully called “a landmark recovery.”
But Alsup wasn’t convinced. He admitted to feeling “uneasy” about how much of the money would actually reach authors, warning that in class actions, members often “get the shaft” once the lawyers take their cut. (Via: Engadget)
What really irked him? The missing details.
The deal didn’t yet specify which works were included, which authors qualified, or even what claim form writers would use to get their slice of the pie.
Basically, it was like promising the world’s biggest buffet but forgetting to hand out plates.
So, Alsup threw the agreement back to the attorneys with homework. He ordered them to provide a full list of works, a full list of authors, and a clear claim process, plus proof that every affected writer gets “very good notice” about what’s going on.
He also demanded guarantees that Anthropic won’t be sued again over the same issue.
Deadlines? September 15 for the works list, and October 10 for the court’s approval of all the missing pieces.
If finalized, the $1.5 billion settlement would dwarf previous copyright deals, setting a precedent for how AI companies pay for gobbling up human-written words. But until then, Alsup isn’t buying it.
He wants assurances that writers won’t just be background characters in a blockbuster legal drama where the real winners are the lawyers.
For now, Anthropic is left waiting. The authors are left waiting.
And Judge Alsup? He’s making sure this deal isn’t just another chapter in the long history of class actions where the people who were wronged end up with the shortest paragraph.
Does Judge Alsup’s rejection of the $1.5 billion settlement show necessary protection for authors’ rights, or will this delay harm writers who need compensation now? Should AI companies be required to get explicit permission before using copyrighted works for training, or are these massive settlements sufficient deterrents for unauthorized use? Tell us below in the comments, or reach us via our Twitter or Facebook.
