{shortcode-934c8cf178cf3180705396096775e6d5eddcaacc}

When Harvard English professor Deidre S. Lynch read an article published in The Atlantic, titled “Search LibGen, the Pirated-Books Database that Meta Used to Train AI”, she learned for the first time that her work was used without consent to train artificial intelligence models.

“I never gave permission, and my publisher never gave permission,” she said.

Lynch is one of the dozens of Harvard professors — and hundreds of thousands of authors — who may be a part of the class in Bartz et al. V. Anthropic PBC, a class-action copyright lawsuit alleging that the company used copyrighted books to train its language learning model.

More than half a million authors are currently a part of the class — though they represent just a portion of all of the authors whose works are in the database.

Sixteen of Lynch’s works were among the millions of pirated texts contained in LibGen, a website that provides free access to scholarly articles and academic books without permission from copyright holders. Other Harvard professors’ works, spanning all departments and divisions, are also included in the database.

The lawsuit was originally filed in August 2024 in the U.S. district court for northern California by a group of authors who alleged that Anthropic, a U.S.-based artificial intelligence startup, obtained pirated works through free digital libraries — like Library Genesis and Pirate Library Mirror — to train its AI model Claude.

The plaintiffs include authors Andrea Bartz, Charles Graber, and Kirk W. Johnson, alongside a larger class of authors.

In June 2025, the court found that Anthropic’s use of legally acquired copyrighted books to train its AI model was protected under the fair use doctrine, which permits limited use of copyrighted material without permission for purposes of criticism, comment, news reporting, teaching, scholarship, and research.

But the court found that Anthropic’s attainment of pirated works, downloaded and stored from online databases, was not protected under fair use.

Following mediation, the plaintiffs and defendants filed a motion for the preliminary approval of a settlement on Sept. 5, which included an agreement from Anthropic that it will destroy its pirated databases and pay $1.5 billion in damages to a group of authors and publishers.

On Sept. 25, a California federal judge granted preliminary approval for a settlement, the largest in the history of copyright cases in the U.S.

Each member of the class will receive a payment of approximately $3,000 per pirated work.

Authors whose works are in the databases are not notified separately, but instead must submit their contact information to receive a formal notice of the class action — meaning a number of authors, including many Harvard professors, may be unaware that their works were pirated by Anthropic.

Lynch said Anthropic’s nonconsensual use of her work undermines the purpose behind why she, and other scholars, write and publish their work.

“All of us at Harvard publish, but we thought when we were publishing that we are doing that — to communicate to other human beings,” she said. “Not to be fed into this mill.”

—Staff writer Victoria D. Rengel can be reached at victoria.rengel@thecrimson.com. Follow her on X @VictoriaRengel_.