Inside AI Policy

February 8, 2025

Meta accused of training Llama AI model with copyrighted content from ‘shadow libraries’

By Rick Weber / January 22, 2025

Facebook owner Meta is being accused of knowingly relying on copyrighted works contained in “shadow libraries” to train its Llama generative artificial intelligence model, according to a third amended complaint in a high-profile lawsuit that could set new legal standards on the data AI developers can use.

“Meta knew these shadow libraries contained pirated works. Several Meta employees sounded the alarm about using what they described as ‘illegal pirated websites’ for training data. Journalists also contacted Meta about its likely...


Log in to access this content.


Not a subscriber? Sign up for 30 days free access to exclusive news and analysis on artificial intelligence regulations and more.