Anthropic agreed to pay $1.5 billion to settle a class action lawsuit by authors and publishers claiming the company trained its large language models on hundreds of thousands of copyrighted books sourced from pirated repositories (Library Genesis and Pirate Library Mirror) without permission or compensation. Under the settlement, Anthropic must destroy original files of pirated datasets and certify which of its released models were trained on pirated material. The settlement allocates roughly $3,100 per eligible work across approximately 448,000 registered works, representing the largest copyright class action settlement in history. For financial institutions, this establishes a quantifiable compliance cost for AI training-data sourcing and creates precedent that AI vendors must document and defend training-data provenance or face material settlement liability.
325 other tracked developments touching the same claims.