Regarding the "real" issue being the use of the API by AI developers: It’s been evident for at least the last few years that this was going on, and I had seen it as a positive thing—that we were helping to create a training corpus that would be freely accessible to everyone, not just a handful of corporations with their own proprietary data.