Particle.news
Download on the App Store

Adobe Sued Over Alleged Use of Pirated Books to Train SlimLM

The case tests whether AI developers can be held liable for training on open datasets that trace back to Books3.

Overview

  • Author Elizabeth Lyon filed a proposed class action in California federal court alleging Adobe used unauthorized copies of her books and others to train its SlimLM models.
  • The complaint says SlimLM was pre‑trained on SlimPajama‑627B, described as a derivative of RedPajama that includes the Books3 collection of roughly 191,000 copyrighted works.
  • Lyon seeks class certification, unspecified monetary damages, injunctive and declaratory relief, and an order to stop use and destroy allegedly infringing copies.
  • Adobe’s own materials state SlimLM was pre‑trained on SlimPajama‑627B, an open dataset released by Cerebras in June 2023 for language model training.
  • The filing adds to widening AI copyright litigation centered on Books3, following actions against major tech firms and Anthropic’s $1.5 billion settlement, as Adobe shares rose about 1.9% on the day reports surfaced.