Firstly, I don't see it as research if it's not published. It's a commercial product if they don't share it and profit from it. If you can reimplement it and publish it, it's yours for the taking.

Secondly, there's so much interesting work outside of large language models.

I don't care too much about what OpenAI get up to. They have a management team trying to become billionaires. That's fine. I'm happy doing science in my living room. Different priorities.