I don't understand why this would be an issue. Firstly, you could just pin your dependencies, but even if you don't, couldn't the default behaviour be to just install the newest scanned version?
Really makes me wonder what the angle here is for them. With open source, I can somewhat understand given market standards become easier to hire for and you get feedback on your tooling. It seems unlikely that either will be true for open-weight models. It also seems unlikely they would be able to establish market domination and then increase prices if everything remains open.
Proponents of free software would agree. In addition, from the programmer's point of view this is usually how things work (unless they own the startup, the equivalent of self-publishing). And for most products that actively gain new users, there is continuous work being put into adding new features and maintenance. So in my mind, this is not a perfect analogy.
I assumed GP was referring to OpenAI, not danswer (given that they mentioned that those companies were training models). And you're still using OpenAI's API, so neither open source and self hosting affect data collection.
What's wrong with "math can't be made illegal"? I think the context for this argument (that I've heard most often) is around e2e encryption, and in that context I agree with it, both from an enforceability and a moral perspective.