Publishers Sue Meta Over AI Training on Pirated Books as Zuckerberg Faces Personal Liability

Five major book publishers and bestselling author Scott Turow filed a class action lawsuit against Meta and CEO Mark Zuckerberg on May 5, 2026, alleging one of the largest copyright violations in history. The suit claims Meta illegally used millions of copyrighted books to train its Llama AI language system. Publishers involved include Hachette Book Group, Macmillan, McGraw Hill, Elsevier, and Cengage. According to court filings, Zuckerberg personally authorized the use of these materials, making him individually liable alongside the company.

The lawsuit alleges Meta obtained these works through piracy websites, including what plaintiffs describe as a systematic campaign to copy protected books without permission or compensation. This represents a significant escalation in the ongoing battle between content creators and AI companies over training data usage rights.

Just weeks before the publisher lawsuit, Meta scored two major victories in related cases. On June 25-26, 2025, U.S. District Judge Vince Chhabria dismissed separate copyright claims from individual authors including Sarah Silverman and Ta-Nehisi Coates. The judge ruled the authors had made "the wrong arguments" and their claims under the Digital Millennium Copyright Act failed because they couldn't prove intentional removal of copyright information.

However, these wins came with explicit limitations. Judge Chhabria emphasized his rulings applied only to the specific cases before him, leaving the door open for future lawsuits with stronger legal arguments. This distinction became crucial when publishers filed their more comprehensive complaint using different legal strategies and evidence.

Meta's legal team has consistently argued that using copyrighted books for AI training constitutes "fair use" under copyright law. The company maintains that training on copyrighted material transforms it into something new and different, serving a different purpose than the original works. Internal documents revealed in court show Meta staff evaluated over 7 million books and deemed them to have no "economic value" for individual licensing, suggesting a calculated decision rather than oversight.

This fair use argument has become the cornerstone of defense strategies across the AI industry, with companies claiming the transformative nature of AI training exempts them from traditional copyright requirements. The publishers' lawsuit directly challenges this interpretation, potentially setting precedent for the entire sector.

The case could fundamentally reshape how AI companies access training data. Publishers are seeking damages that could reach billions given the scale of alleged infringement and the commercial value of AI models trained on their content. The outcome will likely determine whether AI companies must negotiate licensing agreements with content owners or can continue using copyrighted materials without permission.

For publishers, victory could establish a new revenue stream from AI training licenses. For Meta and other AI companies, defeat could force massive retroactive payments and fundamentally alter their business models. The case also raises questions about the viability of current AI development practices if fair use protections prove insufficient.

The lawsuit enters discovery phase in Manhattan federal court, where publishers will seek internal Meta communications about the decision to use copyrighted books. Zuckerberg's alleged personal involvement adds both legal risk and public relations pressure. The case could take 2-3 years to reach trial, with potential appeals extending the timeline further.

Meanwhile, similar cases against OpenAI, Google, and Anthropic continue through courts, creating a patchwork of precedents. Congress may intervene with new AI-specific copyright legislation if courts deliver conflicting rulings. The ultimate resolution will likely come through either a landmark Supreme Court decision or negotiated industry-wide licensing agreements.