A federal judge decided Anthropic did not break copyright laws when the company used millions of books to teach its artificial intelligence system. Judge William Alsup said Monday that training the AI counted as fair use under the law. The ruling helps Anthropic defend against claims from authors who sued the company. Alsup found that using copyrighted materials to train AI systems falls within legal boundaries. The decision marks an important win for AI companies facing similar lawsuits.
Anthropic still faces serious legal trouble over how it obtained the books used for training. The judge said the company must stand trial for allegedly downloading pirated copies of books. Alsup wrote that Anthropic cannot escape responsibility for stealing books just because it later purchased legal copies. The company downloaded at least seven million books that were pirated versions. Court documents show Anthropic knew the books were stolen when it used them.
Three authors brought the lawsuit against Anthropic after discovering their books were used without permission. Andrea Bartz, Charles Graeber and Kirk Wallace Johnson claim the company copied their work from both pirated and purchased sources. None of the authors gave Anthropic permission to use their books for AI training. The company offers an AI service called Claude that can write and read like humans. Courts have not set a date for the upcoming trial about the pirated book downloads.
Anthropic still faces serious legal trouble over how it obtained the books used for training. The judge said the company must stand trial for allegedly downloading pirated copies of books. Alsup wrote that Anthropic cannot escape responsibility for stealing books just because it later purchased legal copies. The company downloaded at least seven million books that were pirated versions. Court documents show Anthropic knew the books were stolen when it used them.
Three authors brought the lawsuit against Anthropic after discovering their books were used without permission. Andrea Bartz, Charles Graeber and Kirk Wallace Johnson claim the company copied their work from both pirated and purchased sources. None of the authors gave Anthropic permission to use their books for AI training. The company offers an AI service called Claude that can write and read like humans. Courts have not set a date for the upcoming trial about the pirated book downloads.
