Senate Grills Meta, AI Giants Over 'Biggest IP Theft in U.S. History'

Testifying before the panel, Maxwell Pritt, a partner at the law firm Boies Schiller Flexner, emphasized the scope of the alleged infringement. “We’re talking about hundreds of terabytes of data — millions of works,” Pritt noted. “Twelve of the books found in the datasets were written by members of this very Senate subcommittee.”

Pritt presented internal documents from Meta and other companies that appeared to show executives were aware that their datasets included pirated content. In one such email, a Meta staffer reportedly said that CEO Mark Zuckerberg personally approved the use of pirated materials. Another document from Anthropic suggested the company used pirated data “to avoid or delay the legal/business slog” of acquiring licensed content.

Authors, academics, and publishers have long raised concerns that their work is being vacuumed up without permission or compensation to power AI systems. One witness, a university professor and published author, testified that several of her books were discovered in AI training datasets without her knowledge. “It’s deeply disturbing,” she said. “We are not being compensated, credited, or even asked.”

Lawmakers from both parties echoed concern. Sen. Richard Blumenthal (D-CT) called for swift legislative action, saying, “We are allowing companies to build empires on stolen work. That must stop.” Some senators floated ideas including mandatory licensing frameworks, statutory damages, or new transparency obligations for AI training.

While the companies named in the hearing have not formally responded to all the allegations, several have previously argued that the data used for training is either publicly available or falls under fair use. However, the subcommittee pushed back hard against that defense, asserting that using entire copyrighted books and proprietary content cannot be justified under current fair use laws.

The hearing signals a potential turning point in how U.S. lawmakers intend to handle the booming AI industry’s opaque data practices. With lawsuits already pending against some of the companies, and increasing bipartisan scrutiny, industry leaders may soon face not only court rulings but sweeping federal legislation regulating how AI is trained.