All Reports

Microsoft moves for summary judgment in Authors Guild copyright lawsuit

Bot Mutiny | September 7, 2026

Microsoft isn't backing down on the legality of training AI on the world's library. On September 4, 2026, the company filed a memorandum of law in the U.S. District Court for the Southern District of New York, moving for summary judgment in the consolidated cases brought by book authors. The filing, first documented by CourtListener, represents a major push to end the litigation by arguing that training large language models (LLMs) is legal as a matter of law. Microsoft is defending its partnership with OpenAI, claiming the use of copyrighted books to build general purpose technology doesn't violate the rights of the creators. The memorandum targets the core of the authors' complaints. Microsoft argues that training an LLM is a transformative use of the underlying material. According to the document, the technology doesn't exist to substitute for the books in their traditional markets. Instead, it uses a varied corpus of data to create something entirely new. ## The fair use defense

Microsoft's legal team is leaning heavily on the four factors of fair use. The filing asserts that the purpose and character of the use is transformative, which should weigh overwhelmingly in their favor. They claim that the authors' arguments regarding "shadow libraries" and "memorization" don't undermine this transformation. The company is asking the court to look past the amount of copyrighted material used. The memorandum states that since the use is transformative, the nature of the underlying works and the specific volume of data used carry little weight. Microsoft also argues there's no evidence of substantial market harm. They're telling the court to reject the plaintiffs' theories about a new AI training market or market dilution. In Microsoft's view, an LLM isn't a replacement for a book. ## Challenging the authors' claims

The filing claims that throughout discovery, the book authors have struggled to substantiate the harms they fear. Microsoft argues that training these models doesn't create a product that competes with books in their traditional markets. The defense cites several recent cases to support its position, including Bartz v. Anthropic PBC and Kadrey v. Meta Platforms. By bringing up these precedents, Microsoft is trying to show a pattern in how courts handle AI training and copyright. The memorandum concludes that the court should grant summary judgment and dismiss the claims. Microsoft is betting that the law will see code and data as more important than the specific creative works used to build them. ## Legal technicalities

This specific document relates to multiple cases under the consolidated heading In Re: OpenAI, Inc. Copyright Infringement Litigation (Case No. 1:25-md-3143-SHS-OTW). It specifically addresses the Books Plaintiffs' Consolidated Cases. The memorandum covers 42 pages of legal theory designed to shield Microsoft and OpenAI from the financial consequences of using massive amounts of data without specific licenses. The filing marks a definitive moment in the battle between the tech companies building these models and the people whose writing makes the models possible.