Recent court documents reveal that Microsoft labeled OpenAI's data scraping practices as 'theft' while both companies allegedly bypassed paywalls to build AI training datasets from The New York Times' content. Internal communications indicate that Microsoft executives recognized the potential harm to publishers and the journalism industry, with concerns about AI models posing an existential threat to their economic viability.
The filings highlight the scale of the data acquisition, showing that OpenAI's datasets included over 91,000 copies of works from various publishers. Microsoft’s CEO testified that any use of paywalled content should be licensed, raising questions about the legality of AI training practices and the implications for copyright law in the tech industry.
Watch for potential shifts in copyright law as this case unfolds, especially regarding AI's use of paywalled content. The implications for publishers and the journalism industry could reshape how AI companies operate and negotiate content licensing in the future.