A file has a name and a path, analogous to conventional storage systems. Putting it in a proprietary system like SharePoint or a Teams channel, to stay in the MS world, has other disadvantages.
I am giving them the benefit of the doubt, but sometimes I think the main goal for these "improvements" is vendor lock in, especially if the topic is around blurring file locations or URLs.
Access rights and relevancy of files is indeed something most alternatives organize better, but content tagging doesn't really work in my experience, because users just don't do it. So the only information about a files content remains the filename and the path.
Since those are fundamental concepts for data organization in most computer systems, it is advantageous to convey these ideas to the user and from my experience this is not that much of a hurdle.
Sure, a network share has disadvantages and without a competent clean up crew it quickly gets convoluted. But that is a problem all other systems have as well. So to improve the current situation, solutions should build upon file systems.
And asking the user if he really wants to save the file is probably just a good idea. If he ignores the message and throws away some work, he has learned a valuable lesson. Users on every knowledge level go through that experience at some point anyway. I think users suffer far more from most proprietary solutions for data exchange that is distinct from copying some files from A to B.
I agree proprietary lock-in is a pain and a huge risk, but I'm talking about fundamental storage models here, not particular implementations (where the problem is lack of interoperability, not the storage model; a proprietary file system doesn't create lock-in, because there are open standards to get your data out of it).
The best content storage model for end users I've seen is the wiki, where you get content versioning for free, although it doesn't separate storage from publication. A wiki where you could keep local separate branches of content would be ideal, but a.f.a.i.k. the only wiki system using that model (shareable local copies) is the Smallest Federated Wiki[1], and it hasn't gained much tracktion.
[1] https://github.com/WardCunningham/Smallest-Federated-Wiki
> And asking the user if he really wants to save the file is probably just a good idea. If he ignores the message and throws away some work, he has learned a valuable lesson
See, that's the kind of comment explaining why users secretly (or openly!) hate computers and think they go after them. Too many developers creating their software are actively disrespecting user's work and their time, even decades after human-computer interaction specialists found the best practices for caring about the user and making their life easier.
A shared folder in a server doesn't separate saving from sharing, but that's a fault of shared folders, not autosave. The popular storage models for end users are utterly broken, but developers don't care because they don't need to endure it, they've got their superior tools that they keep to themselves.