This puts the burden to make sure it's right on the submitter, where it should be. Verification can come at any time after that; the submitter understands the consequences of hallucinated references. Verification can be crowd-sourced (and likely will be).
Nothing stops someone from putting a PDF on the internet. I'm fine with ArXiv holding a high standard.
More than fine, let’s encourage it.
We deserve it, it’s one of the ways to differentiate from the Elsevier et al shitboxes!
Not to mention Zenodo, Academia.edu, etc.