conference - What to do about questions that are designed to embarrass the speaker?

I've attended a lot of seminars and lectures by now, and it's typical for the chair of the seminar to offer audience members a chance to ask questions of the speaker about their work after the presentation.

Most of the time, the questions seek clarification of some aspect of the presentation or focus on a more comprehensive understanding of the research involved.

However, occasionally, I've noticed that questions are purposefully designed to embarrass the speaker. Things along the lines of "That method won't work at all for what you're trying to do. Your results are completely invalid" or "So-and-so's group already did that work years ago. Did you not read their paper?"

Perhaps more disturbingly (I just got back from a really large conference if you can't tell), is that women seem to be more harshly criticized than men, and over trivial issues. For instance, in a few sessions I went to, female graduate students were given really hard times over their presentations while the male grad students were not. All presentations were about the same quality. I guess I'm a little shocked; I'd heard of sexism in academia but hadn't actually seen it (or noticed it) until this conference.

Oddly enough, I've observed that it's typically prestigious professors or researchers that are asking these ostentatious questions. I suppose they figure they have enough "fame" or whatever that their job isn't in jeopardy, and there's no easy way to really prove they're being rude or sexist.

I don't know what's going on here, but it seems to me that these questions, even if they do have technical merit, should be held until after the seminar, where they can be discussed privately with the researcher.

My question here, specifically, is what can be done to minimize these (uncomfortable for everyone) instances? My thoughts are that a session chair should remind the audience to refrain from questions that are accusatory in nature. As a presenter, I'm not sure what can be done in advance to preempt and avoid these questions. Any ideas?