As part of an ongoing copyright lawsuit, The New York Times says it spent 150 hours sifting through OpenAI’s training data looking for potential evidence—only for OpenAI to delete all of its work.