A new lawsuit from a group of writers is putting fresh pressure on the AI industry over how large language models are trained. The complaint names Anthropic, Google, OpenAI, Meta, xAI, and Perplexity, and it centers on an allegation that the companies used pirated copies of books to train their models.
The case includes John Carreyrou, known as the Theranos whistleblower and author of Bad Blood. The writers are not just objecting to copying as an abstract issue. They are arguing that the use of stolen books in model training should carry a real cost for companies whose products generate billions of dollars in revenue.
A familiar dispute enters a new phase
The lawsuit arrives after another group of authors already brought a class action suit against Anthropic over the same acts of copyright infringement. In that case, the judge separated two questions that are often blended together in public debate.
According to the source article, the judge ruled that it was legal for Anthropic and similar AI companies to train on pirated copies of books. But the judge also ruled that it was not legal to pirate the books in the first place.
That distinction matters because it leaves authors with a complicated outcome. The act of model training may be treated differently from the act of obtaining the books. For writers, the practical concern is whether a company can benefit from material that was allegedly acquired through piracy while resolving the matter for an amount that authors see as too low.
Why some writers are dissatisfied
The earlier Anthropic settlement is valued at $1.5 billion. Eligible writers can receive about $3,000 from it. For some authors, that resolution does not answer the core question: whether AI companies are being held accountable for using stolen books to build products that can become extremely valuable.
The new plaintiffs argue that the settlement structure does not reflect the scale of the alleged infringement. Their objection is not limited to compensation per writer. It is also about precedent, leverage, and whether thousands upon thousands of claims can be resolved in a way that the authors believe favors AI companies more than creators.
According to the new lawsuit, the plaintiffs say that the proposed Anthropic settlement seems to serve [the AI companies], not creators.
The filing also says: LLM companies should not be able to so easily extinguish thousands upon thousands of high-value claims at bargain-basement rates, eliding what should be the true cost of their massive willful infringement,
The companies named in the case
The lawsuit names six major AI companies: Anthropic, Google, OpenAI, Meta, xAI, and Perplexity. The shared allegation is that the companies trained models on pirated copies of books.
That common claim is important because it frames the dispute as an industry-wide issue rather than a narrow complaint against one developer. The plaintiffs are targeting the broader practice they say allowed AI companies to build model capabilities from books obtained unlawfully.
The source article does not describe separate allegations against each named company. It presents the case as a collective claim from writers against those six companies, with the central accusation focused on pirated books and model training.
What is at stake for creators and AI firms
For authors, the dispute is about control over their work and the value of that work when it is used in systems that can generate revenue at large scale. The case reflects a concern that compensation tied to a settlement may be far smaller than the value writers believe their books contributed to AI development.
For AI companies, the lawsuit adds to legal pressure around training data. The earlier ruling described in the source article gave AI companies a meaningful point in their favor by treating training on pirated copies as legal. But it also preserved the problem of piracy itself, which is now the focus of renewed litigation.
The result is a narrow but consequential debate. The question is not only whether models can learn from books. It is whether companies can resolve claims over allegedly pirated books at rates that some writers consider inadequate.
The new lawsuit turns that dissatisfaction into a direct challenge. Its message is that authors want the legal system to treat the alleged use of stolen books as a high-value claim, not as a low-cost input that can be settled and moved past.