Internal documents from Microsoft and OpenAI, disclosed during ongoing copyright litigation with news organizations like The New York Times, reveal internal concerns regarding the impact of AI on journalism. Brent Hecht, a director at Microsoft, warned that using news content for AI training could be viewed as massive labor theft and questioned the company's fair use defense.
Microsoft internal data indicates a significant decline in traffic for news outlets involved in the lawsuit, with click-through rates dropping by 51% to 94%. CEO Satya Nadella acknowledged that AI chatbots reduce the need for users to visit original news websites by providing information directly. This phenomenon is described internally as a doom loop where AI diminishes the revenue of news providers.
OpenAI's ChatGPT lead, Nick Turley, noted that using news content to train commercial AI products poses an existential threat to publishers. He characterized chatbots as inherently substitutive, suggesting users have little reason to click original links when information is provided within the AI interface. Concerns were also raised about AI models reproducing copyrighted content.
News organizations allege that Microsoft and OpenAI failed to prevent AI from outputting content highly similar to training data. Furthermore, internal documents suggest Microsoft developed filtering mechanisms that potentially obscure which content is being used for training. Microsoft maintains that its AI products constitute transformative fair use and do not replace news sites.
