In some countries like mine (Iran), you are asked to have at least two papers printed or accepted in an ISI journal before you can claim for your PhD thesis defense.
Publishing papers is not a requirement in many places, but it usually depends on the specific university and graduate program. In the U.S. these things tend to vary from school to school. Sometimes research projects funded by industry prohibit the people involved from directly publishing results.
As for a world record, I'm pretty sure nobody keeps track of that. It is pretty subjective. How would you determine whether or not the paper is from your thesis rather than just being in the same field? There shouldn't be that much value in the number of papers either.
In my dept. (civil engineering), a minimum of 1 published journal article is required for defense. However, my adviser won't schedule any student of his for a defense unless they have published 4-5 journal articles. Keep in mind that other professors in the same dept. go with the 1 paper rule. It varies from a professor to another and from field to field.
Even if there is no formal requirement, here in The Netherlands several Universities/Department in my field require 4 papers to be submitted before Ph.D. defense. Yet, the limit is somehow soft and depends on adviser and impact factor of the journals where the papers are published.
The department of Computer Science of the university I attended in Italy had a requirement of at least 1 journal paper and 1 conference paper published (or accepted) by the end of the PhD. At the time, it was a pretty unique requirement and considered strict, but in recent years more and more Italian universities have implemented a similar policy.
In Sweden the doctoral theses may be in the form of a monograph or in the form of a series of papers presented together with a summary to bind and tie them all ("compilation thesis").
Regarding the compilation theses, it seems, at LTH at least, that it is not strictly required that the papers be published already:
The articles should be of the quality required for publication in international journals.
Nevertheless, I remember distinctly that supervisors expected that, by the time you made your Licentiate dissertation (i.e., about half-way to the doctorate), at least two or three papers be submitted or ready to be submitted to a journal. By the time you are ready to defend the PhD thesis, the submitted papers might be already published or at least accepted, plus maybe two other papers at varying stages.
This, of course, varies and depends on the thesis subject, the study plan, the supervisor criteria, the common practice in your academic Department, even on what's common in the discipline or field of study.
Generally, as stated in the first link, many consider that
there should be more emphasis on the content of what is published than on the number of articles.
At my university, it is expected that there will be at least one publication-worth of work by the submission time, but there is no strict requirement for the paper(s) to be accepted.
For example - I am in my third (hopefully final) year, and I have a joint paper out for conference review and a couple being re-worked after rejection (one journal, one conference).
It may be a little different as I find that computer science tends to weigh conference papers quite highly.
At IIT Bombay (India) it is up to the department and the supervisor to have a certain number of publications. Otherwise, before a year or so even here it was the same scenario of a publication and conference paper.
It is not mandatory, but to have up to date knowledge on current trends in the field of interest is very much needed and that it possible through this medium.
