These pollsters have a reputation to lose. If they publish bogus polls to support a certain media narrative ("orange man dead in the water") and stick with them, they would be exposed as frauds on election day. What I would be looking out for is if the polls shift significantly toward Trump in the week before the election while there are no events which could explain this shift. In this case, I would say some pollsters were indeed tinkering with their polls to push a narrative during the campaign season, and are now reverting their numbers back to where the race really stands so that they can save face. But we're not there yet.
Trump has run an awful campaign, nothing he does or tries seems to work out for him this year, all external events have played into Biden's hands, and he never had any "popularity or goodwill cushion" to fall back on - so the more logical and simple explanation is that he's indeed losing big right now.
I will say though, some outlier polls are clearly inflated. Like that one Quinnipiac poll the other day showing Biden up by 11 in Florida and 13 in Pennsylvania. Even the chair of Democratic PrioritiesUSA super pac called these numbers out as bullshit:
https://mobile.twitter.com/guycecil/status/1313913531834736641?s=21This post was edited by Black XistenZ on Oct 10 2020 06:30am