Adobe Hit With Copyright Suit Over Training AI on Pirated Books

Adobe Inc. willfully infringed hundreds of thousands of copyrighted books by using pirated datasets to download, copy, and store the works to train AI, an author alleged in a proposed class action.

Arthur Kleiner accused Adobe of using the pirated literary works in the “notorious” datasets RedPajama and Common Crawl to train its SlimLM large-language models in a complaint filed Monday in the US District Court for the Northern District of California.

The lawsuit joins a similar copyright suit against Adobe filed by author Elizabeth Lyon in December, and dozens of others against artificial intelligence companies including OpenAI Inc. and ...