Salesforce sued by authors over artificial intelligence software (www.reuters.com)

A group of authors has filed a lawsuit against Salesforce, alleging the company’s artificial‑intelligence software was trained on copyrighted books and other protected works without permission. The complaint seeks to stop Salesforce from using those works to build and deploy its AI features and asks for damages, arguing the company’s training and commercial use of authors’ texts amount to copyright infringement rather than a lawful research activity. The case is significant because it targets one of the largest enterprise AI vendors and could shape how commercial AI systems are built and licensed. At issue are familiar legal fault‑lines for the AI/ML community: whether ingesting copyrighted text to train large language models is fair use, whether model outputs constitute derivative works, and what remedies (deletion of data, injunctions, licensing obligations) courts may impose. A ruling against Salesforce could force vendors and researchers to adopt stricter dataset provenance, opt‑in licensing, or more aggressive filtering, while a defense win could validate current large‑scale scraping practices. Either outcome will influence model training pipelines, compliance costs, and commercial deployment of generative AI across industry.