FWIW, I've run into this with preprints. I had a paper up on a preprint server, and it was under review for a while. In the meantime, I became aware of another paper that pretty clearly used my paper and an earlier closely related one (to the point that some of the language was the same), passing it off as original. When this came up, their response was pretty similar to that in the OP here: things not formally peer reviewed "don't count" and essentially don't exist (and presumably are fair game to plagiarize?).

The paper was eventually published and there's a paper trail, but I wonder how it will all play out.

It feels a little like a wild west to me.