Internal Microsoft and OpenAI documents reveal concerns over news content scraping for AI training
Solo los hechos

Internal Microsoft and OpenAI documents reveal concerns over news content scraping for AI training

Resumen

Leaked internal memos show Microsoft and OpenAI executives acknowledging that large-scale scraping of news articles for AI models could harm publishers and may not qualify as fair use.

Newly unsealed court filings in a lawsuit filed by The New York Times and other publishers contain internal Microsoft and OpenAI documents that discuss the impact of using news content to train large language models. Microsoft Director of Applied Science Brent Hecht described the practice as a significant theft of labor and said it undermined the concept of fair use. OpenAI’s head of ChatGPT, Nick Turley, warned that publishers faced an existential threat from commercial AI products that could replace news sources.

"It is highly unusual that an end-product threatens the economic foundations of its essential suppliers, but that is the situation we have created for our LLM business with respect to its ‘content supply chain,’"

The filings note that Microsoft observed steep declines in click-through rates for several news outlets after the launch of AI tools, with drops ranging from roughly 50 % to over 90 %. Testimony from Microsoft CEO Satya Nadella acknowledged that chatbots can provide information directly, reducing traffic to news sites.

OpenAI internal messages show staff discussing methods to bypass paywalls and noting that users are unlikely to click original articles when the chatbot supplies the needed information. The plaintiffs argue that such substitutive behavior, combined with verbatim excerpts of articles, weakens any fair-use defense.

Microsoft’s spokesperson responded that the company views its AI offerings as transformative and that the cited internal comments represent an individual employee’s view, not corporate policy. Counsel for the New York Daily News said the documents demonstrate that the companies were aware of the potential harm to publishers.

The lawsuit seeks a court ruling that the use of news content for AI training does not qualify as fair use, arguing that without licensing agreements, both news publishers and AI developers risk a “doom loop” in which declining news revenue undermines the supply of reliable information for AI systems.

Sigue leyendo, gratis

Crea una cuenta gratis para seguir las noticias. No se requiere tarjeta.