‘A Moment of Reckoning’: Inside Anthropic’s Landmark Lawsuit

Anthropic, the AI firm behind Claude, is in hot water. 

Right now, the company is preparing for a potentially catastrophic legal challenge that could fundamentally reshape how AI companies train their technologies.

The class action lawsuit centres on allegations that the company downloaded millions of copyrighted works from shadow libraries including LibGen and PiLiMi to train its LLM, Claude. 

Claude has since become one of the world’s most popular Gen AI models, rivalling companies like OpenAI and Google for dominance in the sector.

The details of the case

Federal Judge William Alsup recently delivered a mixed ruling that has divided the AI community.

The judge determined that training AI models on lawfully acquired books qualifies as fair use, eliminating the need for licensing agreements with copyright holders.

Judge Alsup did, however, make an important distinction regarding the acquisition of these materials.

"The judge seems to be suggesting that if you had bought a million books from Amazon in digital form, then you could do the training, and that would be legal, but it's the downloading from the pirate website that is the problem," says Luke McDonagh, an Associate Professor of Law at LSE.

How much is Anthropic being sued for?

The financial implications for Anthropic could be astronomical.

Ed Lee, a Professor of Law at Santa Clara University, has suggested that the ruling could leave Anthropic facing a “business-ending liability”.

Statutory damages range from US$750 to US$150,000 per work, with the upper limit applying if infringement is deemed wilful.

Ed estimates that Anthropic could end up paying damages between US$1bn and US$3bn if just 100,000 works are included in the class action.

In the most extreme scenario, the company could face up to US$1.05 trillion if a jury decides Anthropic wilfully pirated 6 million copyrighted books.

‘A moment of reckoning’

Richard Johnson, Co-Founder and COO of the Data Guardians Network, believes that this case could be era-defining for the AI sector.

“This is a moment of reckoning for the industry,” Richard says.

“For years, everyone knew that AI companies were cutting corners, scraping copyrighted material without consent being the most obvious, but the speed race to beat competitors made it 'acceptable risk'."

The case has achieved class action status, meaning any authors whose works were used without permission can choose to join the lawsuit.

“This turns a handful of plaintiffs into potentially thousands, and means that the damages could run to a huge number – potentially billions of dollars,” Richard explains. 

“Suddenly, the requirement may be on Anthropic to prove that it didn’t use pirated data - or at least that they lawfully acquired it. If this isn’t possible, they are facing a legal and reputational crisis that could last a while.”

Could the Anthropic case create a snowball effect?

Regardless of the verdict, this lawsuit will likely extend beyond Anthropic and set a precedent going forward.

“Many more lawsuits to come in the next 12-18 months,” according to Richard.

“If the court rules that training on pirated data isn't fair use, it could invalidate a huge number of existing models overnight,” he says.

“Public opinion already seems to support the creators - I doubt this will change soon. Such publicity is going to force the AI sector towards a significant rethink - not just on compliance, but on the ethics of data sourcing.”

The political considerations and the future of data sourcing

However, some legal experts suggest that the current political climate could influence the final outcome of the case.

Luke believes that the Trump administration would likely resist any ruling that could “essentially destroy an AI company”, simply because of how fierce the international AI arms race is becoming, with the US and China both vying for superiority in the sector.

The trial, scheduled for 1 December 2025, will be the first certified class action against an AI company over copyrighted materials usage, and its outcome could reshape the sector entirely.

“It’s why the future-forward companies are already looking at ethically sourced frontier data,” Richard explains. 

“Because once the precedent is set, the scramble to rebuild datasets will be brutal and more expensive.”