Publishers Sue Meta Over AI Training: Hachette, Macmillan Lead $Billion Copyright Battle

Five of the world’s largest publishing companies filed a lawsuit against Meta Platforms in Manhattan federal court, accusing the tech giant led by Mark Zuckerberg of defrauding millions of copyrighted works to train its Llama models for artificial intelligence, a development that adds new fuel to one of AI’s defining controversies.

Elsevier, Cengage, Hachette, Macmillan and McGraw Hill, joined by US best-selling author Scott Turow, filed suit on Tuesday alleging that Meta knowingly used copies of books, peer-reviewed scientific journals and novels, including NK Jemisin’s The Fifth Season and Peter Brown’s to train the now-affiliated Wildligen Robot team. AI products.

The complaint, which seeks unspecified damages and class-action status on behalf of dozens of copyright holders, marks the first time academic and commercial publishers have challenged Meta as a unified entity. It also shows the deliberate growth of an industry that has, until now, been largely on the sidelines as writers, newspapers and visual artists fight for their corner.

Maria Pallante, president of the Association of American Publishers, did not mince her words. “Meta-scale breaches are not progress for society, and AI will never be seen well if tech companies prioritize pirate sites over education and imagination,” he said.

Meta has signaled that it will install a strong defense. “AI is enabling the transformation, productivity and creativity of people and companies, and courts have rightly found that training AI with copyrighted material can qualify as fair use,” the spokesperson said. “We will fight this case hard.”

The case opens yet another charge in a battle that is rapidly redrawing the commercial map for content owners on both sides of the Atlantic. Dozens of plaintiffs, from the New York Times, pursuing OpenAI and Microsoft, to a coalition of writers, news outlets and visual artists, have already filed lawsuits against leading AI developers. Legal questions hinge on whether copying copyrighted material to produce new, “adaptive” output qualifies as fair use under American law, and early decisions have been fairly consistent. The first two judges to deal with the issue reached conflicting conclusions last year.

The first big blow came when Anthropic, the AI ​​company backed by Amazon and Google, agreed in 2025 to pay $1.5 billion (£1.18 billion) to settle a class action brought by a group of authors, a sum that would have been found to be many times that amount if the case had gone to trial.

For UK SMEs working in the publishing, marketing, education and creative industries, the results are far from academic. The absence of a coherent licensing regime has left British rights holders exposed to similar allegations, while AI-dependent businesses face increasing uncertainty over which models can be distributed without incurring legal liability.

Benjamin Woollams, CEO of TrueRights, says the industry urgently needs a commercial infrastructure that can match the speed at which AI models are being developed. “All of these cases point to the same basic problem: there is no established way to license creative work and the likes of AI,” he said. “Tech companies aren’t bullies looking for training data, and creators aren’t luddites looking to get paid, but the infrastructure to connect them hasn’t been there yet. This represents a huge opportunity for those in the industry to build a transparent and trusted licensing framework that allows innovation and creator rights to co-exist commercially.”

He points to the influence marketing economy, worth tens of billions of pounds worldwide and built almost entirely on the licensing of rights, as evidence that a commercial template already exists. “Brands and talent are interacting every day on a massive scale. The desire to sell licensed content is there, the economic model is being proven, and creators are becoming increasingly aware of how their IP similarities are used. What’s been missing from AI is a transparent, reliable way to license at the speed and scale these brands need.”

Without such guardrails, Woollams warns, the drumbeat of litigation will grow exponentially. “This kind of conflict and litigation will continue to plague the industry, which will have negative consequences for the kind of collaboration that should power the next generation of creative work, where AI platforms, marketers and talent can build together.”

In Meta, the stakes go beyond the immediate price tag. A successful class certification could expose the group to claims from thousands of copyright holders, while a negative decision could backfire on an industry that has built its competitive edge on a large company’s unrestricted import of copyrighted work. For British SME publishers and creators, this case is a reminder that the rules for dealing with generative AI are still being developed, and that the courts are, in the meantime, shaping the framework.