Anthropic trained Claude on illegally obtained Stanford professors' books

Anthropic illegally trained Claude with Stanford professors’ books AI giant Anthropic illegally scraped and pirated the work of over 10 Stanford professors to train its Claude models, a federal judge ruled in a May 2026 lawsuit. Anthropic previously trained Claude on illegally obtained texts from professors Tobias Wolff, Abraham Verghese, Chang-Rae Lee, Daniel Mason, Hayagreeva Rao and Myra Strober. In Bartz v. Anthropic, which the Authors Alliance calls the largest copyright settlement involving an AI company, Judge William Alsup ruled that Anthropic could no longer pirate books to create a permanent central library. While blocking the piracy of work to train AI, Alsup justified the legal acquisition of books to train Claude’s Large Language Models (LLMs) as fair use. He compared AI training to how humans can read, internalize and remember a book’s themes and style. “Like any reader aspiring to be a writer, Anthropic’s LLMs trained upon works not to race ahead and replicate or supplant them — but to turn a hard corner and create something different,” he wrote. The $1.5 billion settlement went to authors and publishers of nearly 500,000 registered works. Wolff and 27 other authors say the settlement isn’t enough. “To put me to work that way, without remittance, is a form of slave labor,” said Wolff, professor emeritus of English and winner of the National Medal of Arts. “They’re using my work without paying for it. As time goes on, they’re going to make the lives of writers very difficult if there’s no fruit to their labor.” 10 of Wolff’s publications had been used for AI training without permission. “[The settlement] still gives Anthropic pretty free rein to roam through our literature and basically take what they think will be useful and profitable for them,” Wolff said. Wolff, along with English professor Chang-rae Lee, opted out of the Bartz settlement. In May 2026, they filed the separate non-class lawsuit Cruz v. Anthropic, seeking a jury trial for full statutory damages of up to $150,000 per infringed work, surpassing the approximate $3,000 of damages per work offered in the Bartz settlement. Similar to the claims in Bartz, the plaintiffs allege that Anthropic illegally downloaded, reproduced and distributed their copyrighted books from “shadow libraries” like LibGen and PiLiMi, and that Anthropic broke the law by downloading and storing the pirated files in a central library. “Our agents didn’t think the blanket lawsuit covered quite enough of the liability to us and are seeking redress in more particular ways,” Wolff said. If Wolff and his fellow authors are successful, they would gain a permanent injunction against Claude model training using their works. Paul Goldstein, a Stanford Law leading scholar in intellectual property and one of the plaintiffs represented in the class action, acknowledged the difficulty of addressing and preventing copyright issues for AI companies. “The essential problem … is what economists call ‘transaction costs,’” Goldstein wrote in an email to The Daily. “Millions of individual copyrighted works may be copied in the course of any training incident, and it’s simply impracticable for the platform to obtain licenses from the copyright owner of every — or, indeed, any — single work.” Outside of Bartz and Cruz, many authors, visual artists and news outlets, including The New York Times and Getty Images, have filed over 70 infringement lawsuits against AI companies. While Bartz established the current rule for pirating books, the legal future of mass scraping AI models used to train LLMs remains uncertain. The Authors Guild is also pursuing other lawsuits, including one against OpenAI and Microsoft, as part of a “multi-pronged approach to ensure that the law is clear that AI companies must license the books they use,” according to The Authors Guild Chief Executive Officer Mary Rasenberger. “Licensing allows authors and publishers to be compensated and to require the use of filters that ensure their works do not appear in outputs unless separately licensed,” Rasenberger wrote in an email to The Daily. The Authors Guild is working with Congress to create national legislation that will require licensing. “The Guild is up for the challenge,” Rasenberger wrote. “This is something we have been doing for 114 years, and the moment we are in today, while challenging, also presents opportunities for us to firmly realize our mission.” For Wolff, this support is critical for the next generation of writers. “I’m lucky that I’m able to have a life as a writer where I enjoy the profits of my work,” Wolff said. “I worry much more about the young writers I’ve worked with at Stanford and elsewhere being able to do that in the face of these technological challenges, because they’re very hard to grasp and to combat.” Some authors are waging these difficult legal battles to achieve a sense of credit for their work, outside of the settlement payment. “Attribution … is very important to the vast majority of authors, and not just academic authors,” Goldstein wrote. “The great majority of countries have something called the ‘moral right of attribution’ on the books to ensure that authors are credited for their work. The U.S., alone among industrialized nations, has no such moral right, which is a terribly unfortunate situation for authors generally.” As Stanford authors and AI companies continue to navigate copyright challenges for training, both sides must navigate a “foreign land,” Wolff said, meeting at the intersection of creative rights and technological growth. April Yu ('30) is a Columnist for Arts & Life (vol. 270) and a contributor to the News section of The Daily. She is an intended Human Biology and English major from New Jersey.