Publishers sue Meta over alleged use of copyrighted works in AI training
A group of major publishers has filed a lawsuit against Meta Platforms, alleging that its AI models were trained on millions of copyrighted texts without permission.
Several major publishers have filed a lawsuit against Meta Platforms in a US federal court. The plaintiffs include Elsevier, Cengage, Hachette, Macmillan, and McGraw Hill. Author Scott Turow is also part of the complaint.
The case concerns the training of Meta’s Llama models. The publishers allege that millions of books and articles were used without authorisation. These include textbooks, scientific publications, and novels.
The issue is legal and structural. AI systems require large datasets. The question is whether copyrighted material can be used for training without permission. The publishers argue that this constitutes infringement. They are seeking damages and class-action status to represent other rights holders.
Meta disputes the claims. The company states that AI training may qualify as fair use under existing law. It has indicated it will contest the case.
The lawsuit is part of a broader pattern. Authors, media organisations, and artists have filed similar cases against AI companies. These include actions involving OpenAI and Anthropic.
Courts have not yet reached a consistent position. Earlier rulings have differed on whether AI training is transformative use.
