In new legal filings submitted by Microsoft in the copyright lawsuits brought by The New York Times and book authors, the company argued that Copilot rarely reproduces long passages from news articles and books. Microsoft had 8.2 million chat records analyzed, saying they had been selected based on keywords provided to the publishers’ expert and pointing specifically to the use of news sites. According to Microsoft, 59,545 of the records contained at least 16 words in common with news content on which the AI model was based.
The company said that the Center for Investigative Reporting expert identified 51 examples in the data set that “substantially overlapped” with CIR’s work. An expert in the authors’ lawsuit, meanwhile, found only 24 responses containing at least 30 matching words across 8.2 million conversations; matches were found in just 10 of the 212 books examined.
The New York Times disputed Microsoft’s findings, saying the company and OpenAI had been accused of “stealing” its journalism to develop commercial products and needed to be held accountable. The Center for Investigative Reporting and Authors Guild did not immediately respond to requests for comment.
Microsoft argued that the data supported the position that the use of copyrighted content to train AI should qualify as “fair use.” The case is proceeding with the claims brought by publishers and authors against Microsoft and OpenAI consolidated before a single judge. The company requested summary judgment to end the case at an early stage. The Trump administration also submitted a brief this week supporting OpenAI in The New York Times lawsuit.
Why it matters
The files show that the case is not limited to the extent to which AI reproduces copyrighted texts in its outputs; they also concern how the use of this content during the training process should be assessed legally. The statistics submitted by Microsoft indicate that the company is seeking to steer the court toward a fair-use assessment that goes beyond the number of direct similarities. By contrast, The New York Times’ objection maintains that the use of journalistic products to develop commercial AI systems is itself an issue requiring accountability. For this reason, how the court weighs the parties’ differing expert analyses and matching criteria becomes critical. The decision could also determine whether the case proceeds in the copyright dispute between publishers, authors, Microsoft and OpenAI.
Background
Microsoft is not a new name in the FikirPilot archive: over the past 90 days, we have published 4 news reports mentioning this name; the latest was dated September 5, 2026.