Innovation & Tech

Publisher Sues Anthropic and Meta After Rejecting $1.5 Billion Deal

Carlson LLC sued Anthropic and Meta on November 19 over pirated training books, after skipping a $1.5 billion class settlement worth about $3,000 per work. Roughly 135 such suits are now active.

By Olivia Hart

3 min read

Publisher Sues Anthropic and Meta Together After Opting Out of Class Action Settlement
Publisher Sues Anthropic and Meta Together After Opting Out of Class Action SettlementAlicia Christin Gerald / unsplash

What's News

  • Carlson LLC sued Anthropic and Meta on November 19 in N.D. California, case 3:26-cv-14248.
  • Anthropic's Bartz settlement, approved July 2026, set a fund of at least $1.5 billion — about $3,000 per work.
  • Judge Alsup called LLM training 'quintessentially transformative' but faulted pirated sourcing; no precedent set.
  • The Carlson suit joins roughly 135 active U.S. copyright cases against AI companies.
  • DOJ's September brief backing AI training drew protests from 12 House Democrats on November 19.

Carlson LLC filed a federal copyright suit on November 19 against Anthropic PBC and Meta Platforms, accusing both companies of training their AI models on pirated and scanned books without permission. The complaint, case number 3:26-cv-14248 in the U.S. District Court for the Northern District of California, arrives months after the publisher walked away from the largest copyright settlement in U.S. history — a fund Anthropic agreed to cap at no less than $1.5 billion. That works out to roughly $3,000 per work before fees, split evenly between author and publisher. Carlson judged that terms were worth more than a seat in the class.

Why did Carlson opt out of the Bartz settlement?

The Bartz case, filed in August 2024 by a group of authors, alleged Anthropic "pirated" roughly 500,000 books for Claude's training data, drawn from Library Genesis, the Pirate Library Mirror and the Books3 dataset. Internal documents also exposed Project Panama, an effort in which Anthropic bought millions of physical books and cut off their spines for rapid scanning. In September 2025, Anthropic agreed to a settlement fund of at least $1.5 billion. Judge William Alsup approved it in July 2026 — while drawing a line between two things. Sourcing books from pirate libraries violated fair use, he ruled, but "the purpose and character of using copyrighted works to train LLMs to generate new text was quintessentially transformative." Because the case settled, that reasoning binds no one.

Carlson LLC holds the copyrights to works by R. and K. Carlson, including R. Carlson's 1997 bestseller Don't Sweat the Small Stuff... and It's All Small Stuff. The author died in 2006. By opting out, the publisher keeps the right to press its own claims one by one. Kyle Roche of Freedman Normand Friedland says his firm has taken the same route for clients because "it can be difficult to prove that different class members were all affected by AI companies' conduct in the same way." The strategy buys independence; it also buys every cost and risk of litigation alone.

What does the new complaint allege?

The suit claims both Anthropic and Meta harvested training material from the pirate libraries Books3 and Anna's Archive, and acquired additional books by scanning physical copies without permission. The allegations remain unproven. Neither company has yet responded on the merits. Meta's own words may haunt it: internal correspondence unearthed in the separate Kadrey v. Meta Platforms litigation shows the company conceded that clearing rights for training data was "unreasonably expensive" and "incredibly slow." Meta builds the Llama model family; Anthropic builds Claude.

How crowded is the courtroom?

The Carlson suit joins roughly 135 active copyright cases against AI companies in the United States. The federal government has now picked a side. In September, the Department of Justice filed a brief in The New York Times v. Microsoft and OpenAI arguing that "the United States has a strong interest in this court rejecting any argument that training LLMs on copyrighted texts violates copyright law." It was the government's first intervention in an AI-training copyright case. On November 19, twelve Democratic House members sent a letter demanding to know why DOJ acted without consulting the Copyright Office or Patent and Trademark Office, citing "serious questions about potential bias." A separate group of Judiciary Committee Democrats had already criticized the filing in October as contrary to standard practice.

Can one publisher out-negotiate a class?

Opting out shifts the burden onto the rights holder. The same week, Wizards of the Coast — a Bartz participant — announced on November 20 it would give up its publisher share and pass all proceeds, "hundreds of thousands of dollars," to D&D authors. The concession followed author pressure, not any court order. As Justin J. Gunnell and Benjamin J. Shack Sackler note, Bartz showed AI companies "may still face major copyright liability for unlawfully obtained training materials, even when AI model training itself may qualify as fair use." Carlson's dual-defendant suit now tests whether a lone publisher can beat that framework — or set binding precedent that a settlement never could.

By Olivia Hart

Source: epyhia.com

Share this article:

More from Olivia Hart

Olivia Hart

Show full bio

Staff writer covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.

2,535 articles

Related articles

« Previous article