To train the model, OpenAI had to make a copy of NYT's works in order to do it. (Running a scraper to dump websites onto your local storage is making a copy.)

To train the model, OpenAI had to make a copy of NYT's works in order to do it. (Running a scraper to dump websites onto your local storage is making a copy.) NYT's first theory is that the act of copying is a prima facie copyright violation.