That is an interesting question without a good answer. As I understand it (I am far from being an expert), there are sometimes specific exemptions to allow that sharing. However, generally speaking, I think it is a gray area. The journals would surely be shooting themselves in the foot, though, if they tried to sue their contributors. Academics would bring down hell on any journal that tried that. Moreover, it is not clear to me that they would win as it seems to be a customary practice even if it is not an explicitly allowed one. Finally, sharing a single paper might get research or academic exemptions as you aren't copying the whole journal issue. I doubt any publisher wants to go down that road. They are probably better off with the law remaining vague.

Sharing with collaborators may constitute fair use. I wouldn't put one of my own papers on my website, though, for example, for something that isn't published as open access.

They are usually shared under the title of “preprint” or “draft” often, but not always, before the time of actual publishing.

Always seemed like a grey area to me. We didn’t really distribute the copy of the paper with the journal/conference’s name + copyright - though a perhaps a line under the title: “To be published in…”

It depends on the circumstances and the publisher. In many cases publishers permit authors to host the accepted version of the paper (but not the final version that includes revisions based on reviewer feedback) on their personal/institutional website and to email copies of the paper on an individual basis to people who request them.

For example, see https://www.elsevier.com/about/policies/copyright (under "Author Rights") for what Elsevier permits you to do with your own work.

On the other hand, publishers have sometimes filed lawsuits against sites where authors share their papers, e.g. ResearchGate: https://www.nature.com/articles/d41586-018-06945-6

Elsevier have also sent takedown notices to universities where academics have made the final version available on their institutional websites:

https://www.washingtonpost.com/news/the-switch/wp/2013/12/19...

https://osc.universityofcalifornia.edu/2013/12/elsevier-take...

https://news.harvard.edu/gazette/story/newsplus/elsevier-tak...

https://blogs.library.duke.edu/scholcomm/2014/01/28/setting-...

What happens in practice: researchers are usually free to share the draft they had before the publishing process. Which means you can read the text before every sentence was wordsmithed in huge pain to make the paper a quarter page shorter.

> casually share papers with their collaborators, are they technically exposing themselves to lawsuits?

For the final, not open access published papers they certainly do.

No one talks about this, but for many (most?) of the papers they publish, there’s no copyright to assign. When I was in grad school, 100% of my research was federally funded (NSF, Navy, NASA) so they insisted that every product of that research be freely available with no restrictions. When we get an article ready for publication, the Journal would send us the standard “you assign your copyright to us” forms. We’d sign them but also send a copy of “the green form” stating that it was federally funded and we didn’t have any copyright to assign. They just ignore that.