*

Anthropic faced potential liabilities in the billions due to piracy ruling

*

OpenAI, Microsoft ( MSFT ), Meta battling similar lawsuits

*

Fair use defense remains contentious in AI copyright cases

By Blake Brittain

Aug 27 (Reuters) - Anthropic's class action settlement with a group of U.S. authors this

week was a first, but legal experts said the case's distinct qualities complicate the deal's

potential influence on a wave of ongoing copyright lawsuits against other

artificial-intelligence focused companies like OpenAI, Microsoft ( MSFT ) and Meta Platforms ( META )

.

Amazon ( AMZN )-backed Anthropic was under particular pressure, with a trial looming in

December after a judge found it liable for pirating millions of copyrighted books. The terms of

the settlement, which require a judge's approval, are not yet public. And U.S. courts have just

begun to wrestle with novel copyright questions related to generative AI, which could prompt

other defendants to hold out for favorable rulings.

Anthropic was in "a unique situation," said Cornell Law School professor James Grimmelmann,

with as much as $1 trillion in piracy damages at stake in its worst-case scenario.

"It's possible that this settlement could be a model for other cases, but it really depends

on the details," he said.

The authors in Anthropic's case accused the AI company of using millions of pirated books

without permission or compensation to teach its AI assistant Claude to respond to human

prompts.

Anthropic, like other AI copyright defendants, countered that its actions were legal under

the doctrine of fair use, which allows the use of copyrighted works without permission in some

circumstances.

U.S. District Judge William Alsup in San Francisco ruled in June that Anthropic made fair

use of the authors' work to train its AI, but said the company violated copyright law by saving

pirated books to a "central library" that would not necessarily be used for AI training. That

created potential liabilities of billions of dollars for Anthropic, which faced an upcoming

trial in December.

The two sides told the court on Tuesday that they had settled the case in principle, the

first accord reached in a copyright lawsuit over generative AI training. Alsup, who must approve

the settlement, ordered the parties to submit details to the court by Sept. 5.

Chris Buccafusco, a law professor at Duke University, said he was surprised Anthropic chose

to settle. Anthropic was "in a position of decent strength" because of Alsup's fair-use

determination, Buccafusco said, despite the piracy decision.

"Given their willingness to settle, you have to imagine the dollar signs are flashing in the

eyes of plaintiffs' lawyers around the country," he said.

Anthropic and an attorney for the authors did not immediately respond to requests for

comment on Wednesday. Universal Music Group, which has separately sued Anthropic over

its alleged misuse of song lyrics to train Claude, also did not respond to a request for

comment.

OpenAI, Meta and Microsoft ( MSFT ) did not respond to questions about how the Anthropic settlement

could shape their ongoing AI litigation.

FAIR USE QUESTIONS

The fate of the pending generative AI lawsuits could hinge on fair use, a still-evolving

concept that no court had addressed in the cases until June.

Grimmelmann said Anthropic's settlement removes an early opportunity for a federal appeals

court to consider fair use and issue a decision that would be binding on other cases and likely

tee up the issue for the U.S. Supreme Court.

Two days after Alsup's ruling on fair use and piracy, another judge in the same San

Francisco court took a different approach in a similar author lawsuit against Meta. The decision

by U.S. District Judge Vince Chhabria mostly ignored piracy issues like those Alsup addressed,

but found that Meta's conduct may not be protected because its AI could be used to "flood the

market" with replacements for the authors' work.

Chhabria ruled for Meta on fair use but said he did so only because "these plaintiffs made

the wrong arguments." Alsup, meanwhile, downplayed fears of market displacement in the Anthropic

case.

An attorney for the authors suing Meta declined to comment on Anthropic's settlement.

Reuters News has licensed its content to Meta for AI use, and its parent company Thomson

Reuters ( TMSOF ) is involved in a lawsuit against Ross Intelligence that argues Ross misused

copyrighted material to train an AI-powered legal search engine.

Decisions on fair use in dozens of other AI cases are unlikely before next year. Given the

stakes, unpredictability surrounding the rulings could provide an impetus to settle, experts

said. But it could also encourage them to hold out in hopes of a sweeping win like Google

obtained from the Supreme Court in 2016 in a fair-use dispute over its Google Books

project.

"The one thing that was clearly going to help was an across-the-board, as-a-matter-of-law

fair use ruling" in the Anthropic case, Buccafusco said. "That would have been the real solution

for all of the AI platforms."