Senate Judiciary Subcommittee Hears Allegations That AI Firms Trained Models on Pirated Books

At a hearing of the Senate Judiciary Committee’s Subcommittee on Crime and Counterterrorism, Chairman Hawley and Ranking Member Durbin presided as witnesses testified that major artificial-intelligence companies used pirated copies of books and scholarly works to train large language models.

The testimony, during the hearing titled “Too Big to Prosecute,” focused on allegations that companies including Meta and Anthropic obtained copyrighted works from illicit online repositories and peer-to-peer “torrent” networks rather than licensing material from copyright holders.

The witnesses said the scale of the alleged activity is substantial and legal accountability is unsettled. Max Pritt, a lawyer representing authors in litigation against Meta, told the panel that documents in that case show Meta acquired more than 200 terabytes of copyrighted books and articles and distributed over 40 terabytes of material via peer-to-peer networks. “This is likely the largest infringement of American intellectual property by U.S. companies in our nation’s history,” Pritt said.

Why it matters: witnesses and senators said the issue could undercut creators’ incomes, distort marketplaces for licensed…