Publishers sue over AI training
Chicken Soup for the Soul and other publishers filed suit alleging major tech firms ingested copyrighted works into AI training datasets—part of a growing wave of litigation over model training sources. The case adds legal complexity to large‑scale model use and distribution decisions. (x.com)
Chicken Soup for the Soul LLC filed its complaint on March 17, 2026 in the U.S. District Court for the Northern District of California as case No. 5:26‑cv‑02333. (law.com) The filing names Apple, Google, Nvidia, Meta Platforms, OpenAI, Anthropic, Perplexity AI and Elon Musk’s xAI as defendants. (theoutpost.ai) The complaint alleges defendants sourced bootleg copies from “shadow libraries,” citing repositories such as The Pile, LibGen, Z‑Library and Anna’s Archive as training inputs. (lunch.publishersmarketplace.com) Plaintiff counsel listed on the filing include Freedman Normand Friedland and Stris & Maher, with partners Kyle Roche, Velvel Freedman and Elizabeth Brannen among the attorneys of record. (theoutpost.ai) The complaint advances a factual allegation that the generative‑AI industry was seeded when a single OpenAI employee first downloaded pirated books in 2018, a claim the complaint uses to trace later model training practices. (publishersweekly.com) The filing emphasizes that Chicken Soup’s first‑person, emotionally‑toned narratives are especially valuable for model training and asserts defendants built “systems now worth many hundreds of billions of dollars” without licensing those works. (theoutpost.ai) Anthropic, named as a defendant in this suit, previously reached a proposed settlement to pay at least $1.5 billion to resolve a separate authors’ copyright class action in 2025. (news.bloomberglaw.com)
