Somehow the name suggested to me that I could find ways to reproduce the model. I understand that it is huge and probably undoable for a single person without large amounts of resources, but it still annoys me that the term "open" is (ab)used in this way.
One could say they are being cautious since It's a very novel technology open for misuse.
And profit. You don't just release complex, highly capable models for charity. They are a for-profit company after all. It is a business they have the right to use their edge for profit.
Also they are not as closed. They offer APIs for many products and some other ones outright release their sources (especially whisper, a godsent gift for me). Unlike Google's AI research which one commenter pointed out.