Skip links

Microsoft Reports Minimal Use of NYT Articles via Chatbot Access

Microsoft has responded to ongoing copyright claims from major publications such as The New York Times and various book authors, asserting that its AI tool, Copilot, seldom reproduces complete sentences from news articles or books. This statement emerged from recent legal filings as the tech giant seeks to defend its practices against allegations of copyright infringement.

In the context of the legal proceedings, Microsoft disclosed over 8.2 million chat logs from Copilot to an expert working for the publishers involved in the lawsuit. The logs were selected for their relevance, particularly focusing on keywords associated with the publishers’ websites, as they were deemed most likely to showcase the use of these organizations’ materials. According to Microsoft’s findings, 59,545 of these chats contained at least 16 words in common with news content that contributed to training the AI model. An expert from the Center for Investigative Reporting identified 51 instances of “substantial overlap” regarding its content. Furthermore, another expert working on behalf of the authors concluded that only 24 responses out of the 8.2 million chats featured at least 30 matching words from the texts. Only 10 of the 212 books reviewed revealed any similarities, according to Microsoft’s claims.

Microsoft contends that these numbers strengthen its argument in favor of considering the use of copyrighted content for AI training as fair use. The company maintains that although tools like Copilot utilize copyrighted material, their final applications serve significantly different purposes compared to the original content. Microsoft concludes that any occurrences of text reproduction do not detract from the transformative nature of large language model (LLM) training.

The legal battle involves a coalition of publishers and authors who allege that both Microsoft and OpenAI have developed products based on their works and are now competing against them by redistributing copyrighted content. To facilitate the legal process, the claims from publishers and authors have been consolidated under a single judge, despite pushback from the plaintiffs. Microsoft filed its latest documents on Friday, pushing for a summary judgment that could bring the case to an early closure. Notably, the Trump Administration recently expressed its stance in support of OpenAI in the New York Times case.

Editor’s Take

This legal skirmish highlights the evolving landscape of copyright and AI integration. As technologies like Microsoft’s Copilot become more prevalent, the discussions surrounding fair use and the boundaries of turf between creative works and AI-generated content will intensify. For users and businesses alike, the outcomes of such cases could redefine content usage and AI development strategies.

Source: www.theverge.com

Leave a comment