If the conference rules forbid posting to the arXiv, the correct course of action is to withdraw the conference submission and to boycott that particular conference.

@Arno Even though you are suggesting the correct way of approaching this issue if my colleague was a professor or something, but as Massimo mentioned, he is definitely not in such a position. This is his first experience of publishing a paper, and if he fails the fight (for whatever reason), it would negatively affect his career.

Note: arXiv is unlikely to actually take down the paper; what you can do is to "withdraw" it, which merely means updating it to an empty file (while the existing preprint still is available as an old version). I'm not sure if the conference chair will be happy with this. My personal suspicion is that the conference chair is too incompetent to correctly interpret the output of their plagiarism detection software; the easiest way to deal with this might be (1) to try to explain things to the chair, and if this fails, (2) to move up the org chart (the series in which the proceedings are ...

... getting published should have an editorial board!).

@darijgrinberg I suspect that it's not about plagiarism but about having the exclusivity of the submission.

@MassimoOrtolano: Maybe. In either case, talking to the chair and then escalating to the editors of the series sounds like a reasonable approach to me.

@MassimoOrtolano The chair was clear in the email (I'm quoting): "This violates IEEE rules." ... so, I agree with darijgrinberg! My colleague followed your suggestion and withdrawn the paper from arxiv (it's not an immediate procedure though and arxiv scheduled the withdrawal for August 3). He also sent an email to the chair for clarification and I hope he doesn't have to go into the fight or escalate the problem to IEEE. I will share their reply here as soon as I receive one...

The general chair still believes it’s against the law without further explanation. Paper is withdrawn, and IEEE has been notified.

@MassimoOrtolano ArXiv rejected my collegues request for withdrawing the paper! They replied "Please note that having a paper under review or newly published is not a sufficient reason for withdrawal, as previous version(s) will still be available to users. arXiv is an electronic repository for research papers, and announced papers are meant to be available in perpetuity. The license applied by the submitter to the work cannot be revoked. As a result, you request has been denied". My colleague is now waiting for IEEE response to see what they can do to fix this issue.

@MichelGokanKhan That is very unfortunate. I wish IEEE will give a positive response to your colleague and contact the conference chair. Personally, this kind of issue is one of the reasons I've been always very cautious in posting preprints on arXiv and for which I eventually stopped altogether (I know, it's not nice but one should weigh pros and cons).

@MassimoOrtolano Thanks for the comment. To be honest, I disagree. Publishing in credible venus/journals takes months, if not years and I still believe publishing preprints somewhere like arXiv is extremely beneficial for both the research community and the researcher. We have all the rights to do so and the stupidity of an incompetent chair shouldn't stop us from doing the right thing. That's just my view, but you probably have much more experience than me :-)

@MichelGokanKhan Indeed I do agree that ideally publishing preprints is beneficial. In practice, unfortunately indeed, the drawbacks in some cases can outweigh the benefits. For instance, another issue that I found about preprints is that then many people cite the preprint and not the published paper. Since in my country promotions are strictly associated to citation counts of the published papers, it turns out that it's not in my best interest to submit preprints. It's a crappy system, but it's the one I have to live in (there are protests against such a system but it's a long way to go).

@MassimoOrtolano, I see. My personal experience was that if someone cites the preprinted version in arXiv, at least Google scholar is smart enough to count that citation for the original published paper (if it has the same title and authors (and probably abstract?)). Actually that's one of the benefits of publishing preprints: to gain citations. But if that's not the case and citing the preprinted version is not always counted, then I think it's a major drawback for publishing preprints. I didn't encounter this issue yet (or heard of it), but now that you mentioned it, I'll investigate this...

@MichelGokanKhan All's well that ends well! :)

@MichelGokanKhan - where is the update with the IEEE email? I cannot see any answer from you on this question, and I cannot find that update...

@Bogd What matters is that they let the conference chair know that it is not against the rules and regulations to publish a preprint in arXiv with appropriate copyright notes. The conference chair is convinced now. The email itself doesn't really matter :-)