Textbook authors sue OpenAI and Microsoft over AI training

Six textbook authors, including Boston College finance professor Alan Marcus, have filed a proposed class action against OpenAI and Microsoft in the US District Court for the Southern District of New York. According to The Heights’ report, the complaint filed on August 14, 2026 alleges that OpenAI downloaded copyrighted books from “shadow libraries,” placed them in datasets and computer memory, and converted them for training its language models and ChatGPT. The allegations focus on the Books1 and Books2 datasets, which allegedly contain ebooks obtained from Library Genesis. The plaintiffs also claim that OpenAI developed software to remove copyright notices, titles, ISBNs and other copyright-management information.

The complaint attributes to Microsoft a role in providing training infrastructure and incorporating the models into products such as Bing Chat and Copilot. The authors also allege that model responses can provide substitutes for textbooks and reduce sales; these claims have not been adjudicated. On September 16, 2026, Judge Sidney Stein reportedly stayed the action pending decisions on competing summary-judgment motions in related litigation. In a brief, the US Department of Justice argued that training AI models on copyrighted writing is fair use, representing the government’s position rather than a judicial ruling. At publication, the court filings cited by the article indicated that OpenAI and Microsoft had not responded to the allegations in this lawsuit.