Microsoft reports minimal interest in NYT articles via its chatbot.
Microsoft’s Copilot seldom reproduces complete sentences from news articles or books, let alone significant portions that could replace the original content, according to the company’s recent legal filings as it contests copyright claims from various publishers and authors.
In the course of the lawsuit’s discovery phase, Microsoft shared 8.2 million Copilot chat logs with an expert engaged by the news publishers. These logs were selected specifically for containing keywords related to the publishers’ websites, making them the most likely to include their works. The analysis revealed that 59,545 of these logs had at least 16 words in common with the news content that informed the AI model. An expert from the Center for Investigative Reporting identified 51 instances of “substantial overlap” with their work within the dataset. Additionally, an expert involved in the authors’ lawsuit found that out of the 8.2 million interactions with Copilot, only 24 responses featured at least 30 matching words. Microsoft claims that only 10 of the 212 evaluated books showed any matches.
The New York Times disagrees with Microsoft’s interpretation of the findings. “The documents and testimony uncovered during discovery lead to only one conclusion: Microsoft and OpenAI stole from The New York Times to create commercial products that substitute for its journalism, threaten its business, and undermine its industry,” stated Ian Crosby, the Times’ lead counsel. “We look forward to Microsoft and OpenAI being held accountable for their theft.” The Center for Investigative Reporting and the Authors Guild did not respond to requests for comment.
Microsoft contends that these figures support its argument that using copyrighted content for AI training datasets should be classified as fair use. While systems like Copilot utilize copyrighted material, Microsoft asserts that the resulting applications serve purposes that are significantly different from the originals. The occasional reproduction of text sections, it argues, “hardly undermines the transformative purpose of LLM training.”
These filings are part of a legal dispute initiated by publishers and authors against Microsoft and OpenAI, who the plaintiffs allege have developed products based on their works and now compete directly with them by reusing copyrighted content. The claims from news publishers and book authors have been consolidated under a single judge to streamline proceedings, despite objections from the plaintiffs. Microsoft submitted its filing on Friday, seeking a summary judgment to potentially resolve the case at an early stage. (The Trump administration also expressed interest in the New York Times case this week, supporting OpenAI.) If the judge rules in favor of the publishers and authors, the legal battle will proceed in court.
