Microsoft has told a US court that its Copilot chatbot only rarely reproduces substantial chunks of copyrighted news articles or books, as it seeks summary judgment in a lawsuit brought by The New York Times, book authors and other publishers. The filing, submitted on Friday, forms part of Microsoft and OpenAI's defence against claims that they built commercial products on copyrighted work and now compete directly with publishers by regurgitating their content. Microsoft argues the low rate of overlap supports its case that training AI models on copyrighted material should count as fair use, since the resulting systems serve a "transformative" purpose distinct from the original works.

As part of discovery, Microsoft handed over 8.2 million Copilot chat logs, selected because they were likely to reference the plaintiffs' content. Of these, 59,545 contained at least 16 matching words with news material used to ground the model, while an expert for the Center for Investigative Reporting found only 51 instances of "substantial overlap." In the authors' case, just 24 of the 8.2 million conversations had 30 or more matching words, and only 10 of 212 books examined showed any matches at all. The case, which could end at this stage if the judge grants summary judgment, has also drawn a supporting statement from the Trump administration backing OpenAI.