Five major book publishers and novelist Scott Turow have filed a class-action lawsuit against Meta and Mark Zuckerberg personally, alleging the company “engaged in one of the most massive infringements of copyrighted materials in history” to train its Llama AI models. The lawsuit, filed May 5, 2026, accuses Meta of knowingly ripping copyrighted works from notorious pirate sites and feeding them into its AI training pipeline.

This isn’t the Sarah Silverman playbook. It’s different — and potentially more dangerous for Meta.

What Makes This Lawsuit Different

Previous copyright cases against Meta focused on whether training AI on copyrighted material constitutes fair use. A federal judge ruled for Meta last year on that exact question, though he pointedly noted the ruling “does not stand for the proposition that Meta’s use of copyrighted materials to train its language models is lawful.”

This new lawsuit goes further in three critical ways:

1. Zuckerberg is named personally. The lawsuit doesn’t just name Meta as a corporate entity — it names Mark Zuckerberg as an individual defendant. That’s unusual and intentional. The plaintiffs allege Zuckerberg personally ordered and directed the copyright infringement, making it harder for Meta to shield him behind corporate liability.

2. Circumvention of DRM. The suit alleges Meta didn’t just scrape publicly available material — it knowingly circumvented digital rights management (DRM) protections to access copyrighted works. If proven, this brings in the Digital Millennium Copyright Act (DMCA), which carries substantially stiffer penalties than standard copyright claims.

3. The pirate sites connection. The plaintiffs point to Meta’s use of materials from LibGen, Anna’s Archive, Sci-Hub, and Sci-Mag — sites that exist almost exclusively to distribute pirated academic and literary content. Internal Meta communications, revealed in earlier lawsuits, showed employees discussing how to handle “media coverage suggesting we have used a dataset we know to be pirated.”

The “Word-for-Word” Problem

The plaintiffs’ most concrete allegation: when prompted with just two sentences from Cengage’s bestselling textbook Calculus: Early Transcendentals, 9th Edition, Llama begins reproducing the continuation word-for-word. That’s not “learning patterns” — that’s memorization and regurgitation.

This is the smoking gun that the fair use defense struggles with. AI companies have long argued that training on copyrighted material is no different from a human reading a book to learn. But humans don’t recite textbooks verbatim on demand. Models that memorize and reproduce training data look a lot more like digital photocopiers than attentive students.

The Context: Publishing Industry Already Bleeding

The publishing industry has been watching the AI copyright wars with existential dread. Book sales are down. The rise of AI-generated summaries and content threatens to decimate what remains of the market. And publishers have seen what happened to the news industry, where AI summarisation tools are already cannibalising traffic.

At the same time, the precedent is mixed. Authors won a $1.5 billion settlement from Anthropic last year. But a separate author lawsuit against Meta was largely dismissed. And against that backdrop, Anthropic secured a ruling that training on legally purchased books without permission is fair use — though the same judge allowed claims about pirated material to proceed.

This new suit tries to split the difference: acknowledge fair use might apply to legally obtained content, but argue that Meta crossed the line when it went to pirate sites and deliberately circumvented DRM.

What This Means for NZ

The New Zealand publishing industry — small, independent, and disproportionately important to our culture — should be watching this closely. NZ publishers don’t have the resources of Macmillan or Hachette to litigate. If this suit clarifies that pirated training data is actionable, it creates a framework for local authors to protect their work. If it fails, the message is clear: if Meta can take your book without paying, so can any AI company, and you have no recourse.

The NZ Copyright Act is under review, and this case will inform how our own lawmakers think about AI training exemptions. NZ authors like Eleanor Catton, Witi Ihimaera, and Becky Manawatu have work that would be enormously valuable for training literary AI models. Should they be compensated? This case will help answer that.

🔍 THE BOTTOM LINE

This isn’t business as usual in AI copyright litigation — naming Zuckerberg personally, alleging DRM circumvention, and pointing to pirate sources makes this the most serious copyright challenge Meta has faced, and a potential turning point for how we treat AI training data.

❓ Frequently Asked Questions

Q: Could this affect Kiwi authors and publishers? Yes. NZ authors and small publishers lack the resources to sue tech giants. If this case establishes a legal precedent that AI training on pirated material is infringement, it protects NZ content creators. If it fails, it signals that copyright has little teeth against AI.

Q: What’s the difference between this case and earlier AI copyright lawsuits? Earlier cases tested whether training AI on copyrighted content is “fair use.” This case focuses on how Meta got the content — from pirate sites with DRM circumvention — and names Zuckerberg as an individual defendant. It’s an entirely different legal theory.

Q: What happens if Meta loses? Meta could face significant damages — potentially billions — plus be ordered to destroy models trained on infringing data. The publishing industry would gain leverage for licensing deals. The broader AI industry would need to fundamentally retool its data sourcing.

Q: Doesn’t fair use cover this? Fair use is a defence, not a guarantee. Courts have been split. The key question is whether knowingly downloading content from pirate sites and stripping DRM protections crosses the line from “fair use” into “theft.” Many legal experts think it does.

📰 SOURCES

• The Verge — “Book publishers sue Meta over AI’s ‘word-for-word’ copying”
• NPR — “Scott Turow, Macmillan, McGraw Hill sue Meta for AI copyright infringement”
• The Washington Post — “Publishers sue Meta over training of AI platform Llama”
• CBS News — “Publishers, author Scott Turow accuse Meta and Mark Zuckerberg of training AI on copyrighted works”
• The New York Times — initial report on the lawsuit filing