I’ve come across a somewhat bizarre case where CrossRef DOIs appear to have been abandoned by a publisher, and the destination of those DOIs has been taken over by a site that seems to be a content farm.
The journal FRAGMENTA ENTOMOLOGICA now has DataCite DOIs for its articles, e.g. AFROTROPICAL TYCHIUS: DESCRIPTION OF FIVE NEW SPECIES AND DESIGNATION OF A NEOTYPE (Coleoptera, Curculionidae) | Fragmenta entomologica. But at one point this journal had CrossRef DOIs, such as 10.4081/fe.2013.24 which for which metadata is still in CrossRef’s search engine: Crossref Metadata Search
The DOI 10.4081/fe.2013.24 resolves to http://www.fragmentaentomol.org/index.php/fragmenta/article/view/24 which redirects to https://www.fragmentaentomol.org (screen shot below).
It looks like the publisher abandoned both the DOIs and the original web site, someone has taken over that domain, and now a bunch of CrossRef DOIs no longer point to scientific articles but a bunch of spam. Meanwhile the articles now live on a different site complete with different DOIs. I don’t think the publisher really understood the point about persistent identifiers.