A judge just called AI training legal — and slapped Anthropic with a $1.5B fine
a judge ruled ai training is legal but fined anthropic $1.5b for pirated books, signaling a split in how courts view training data versus data sourcing.
in a ruling involving anthropic, judge william alsup distinguished between the act of training models—which he likened to a human reader studying literature—and the act of copying protected works. while he found the training itself lawful and analogous to fair use, he levied a $1.5b fine specifically because the books were sourced from pirated shadow libraries. this creates a legal precedent that ai training is likely permissible under current copyright laws, provided companies avoid datasets with blatant copyright status issues. the decision highlights the growing gap between existing statutes, last updated in 1976, and modern generative ai capabilities.