Microsoft says virtually nobody was grabbing NYT articles through its chatbot
Key Points:
- Microsoft states in legal filings that its Copilot AI rarely reproduces full sentences or substantial content from news articles and books, countering copyright claims from publishers including The New York Times.
- In discovery, Microsoft provided 8.2 million Copilot chat logs, finding only a small fraction contained overlapping text with copyrighted news or books, with minimal instances of substantial overlap identified by experts.
- The New York Times disputes Microsoft's findings, accusing Microsoft and OpenAI of stealing content to create competing commercial products that harm the journalism industry.
- Microsoft argues that using copyrighted material for AI training constitutes fair use, emphasizing the transformative nature of AI systems like Copilot compared to the original works.
- The legal battle between publishers, authors, Microsoft, and OpenAI continues, with Microsoft seeking summary judgment to end the case early, while the Trump administration has filed a statement supporting OpenAI.