AI copyright news, graded by source

Authors ask the court to deny Microsoft summary judgment, arguing it is directly liable for passing copies of their books to OpenAI and vicariously liable for OpenAI's copying

In a heavily redacted 49-page brief filed Oct. 24, the class plaintiffs (authors and the Authors Guild, asserting 194 works) oppose Microsoft's motion for summary judgment. They argue Microsoft directly infringed by distributing copies of the works to OpenAI in exchange for valuable consideration, a use they say must be judged on its own rather than folded into model training, and that handing over full, unchanged copies is not transformative. They also argue Microsoft is vicariously and contributorily liable for OpenAI's infringement and that training a large language model on copyrighted books is not fair use. The brief points to Microsoft's 2024 and 2025 proxy statements, which say it does not train on data from domains on the U.S. Trade Representative's Notorious Markets list, a list that has long included LibGen, and says Microsoft witnesses could not recall the defendants' reliance on LibGen. It cites the Third Circuit's Sept. 29, 2026 ruling in Thomson Reuters v. ROSS. Much of the factual section is blacked out; the unredacted brief is under seal. These are the authors' claims, not findings.