In CS, it's widely accepted that you can publish your "drafts" publicly. Basically just the real deal without the notices of publication (I guess maybe legally you have to leave some differences in? I never bothered, and I doubt the publishers who everyone kinda hates would dare creating this kind of controversy).

As others have pointed out, some countries mandate national or institutional repositories.

A good grassroots way out of this is educating researchers to always take one of these options. Let publishers overplay their hand and get crushed.

Of course, it would be nicer if regulators realized how ridiculous this all is and crushed them without the need for public outcry, but one may dream.