The question has been answered by Greg Kuperberg: there has been no significant changes in the arXiv management. An "insignificant" change is the use of an overlap software which, as I understand now, triggered the whole problem. I think that the use of that software is potentially dangerous both for the authors (who may be falsely accused in plagiarism or in illegally multiplying their number of publications) and for the arXiv (because of potential lawsuits). Several other people expressed the same concern and I hope the people in charge of the arXiv will listen. Other than that, the question can be safely closed.

I left a comment on the original post voting that the question remain open. But as it is now answered, I am perfectly happy to have it stay closed as no longer relevant. I do think that arXiv is absolutely by far the most valuable research tool for most mathematicians (if NSF were to fund it at a fraction of what they fund colliders, which don't produce nearly as much research, then everything would be hunky dory). I agree with those comments that the original post was unfortunately written: too accusatory and complainy.