A group of writers has filed a lawsuit against several major artificial intelligence companies, alleging their models were trained using pirated copies of copyrighted books, and challenging how earlier legal settlements have addressed responsibility for that practice.
Lawsuit Targets Multiple AI Companies
The plaintiffs include journalists and authors such as John Carreyrou, known for his reporting on Theranos and the book Bad Blood. The lawsuit names Anthropic, Google, OpenAI, Meta, xAI, and Perplexity as defendants.
According to the filing, the writers accuse the companies of training large language models using unauthorized digital copies of their books. The complaint focuses on the alleged use of pirated material rather than the output produced by the models.
Connection To Earlier Anthropic Case
The lawsuit follows a previous class action case brought by a separate group of authors against Anthropic over similar allegations. In that case, a judge ruled that training AI models on pirated books was legal under existing interpretations of the law, while also stating that the act of pirating the books itself was not legal.
That earlier case resulted in a settlement valued at $1.5 billion. Under the terms of that agreement, eligible writers were expected to receive payments of about $3,000 each.
Objections To Settlement Terms
Some authors expressed dissatisfaction with the outcome of the Anthropic settlement. According to the new lawsuit, the plaintiffs argue that the agreement did not hold AI companies accountable for acquiring and using unauthorized copies of books to train systems that generate significant revenue.
The filing states that the proposed settlement “seems to serve [the AI companies], not creators.” It further argues that the agreement allows companies developing large language models to resolve a large number of copyright claims at what the plaintiffs describe as low compensation levels, without addressing the scale of the alleged infringement.
Claims Raised In The New Filing
The lawsuit asserts that companies building large language models should face responsibility for the underlying use of pirated works, not only for downstream uses of the models. It claims that the current legal approach allows companies to avoid what the plaintiffs describe as the true cost of large-scale copyright infringement.
The defendants have not responded publicly to the specific allegations raised in the new lawsuit at the time of filing.
Featured image credits: Wikiemedia Commons
For more stories like it, click the +Follow button at the top of this page to follow us.