Off the cuff:
as microformats diffuse throughout the web, malevolent uses will inevitably rise.
if you imagine that microformats allow you to use the web as a database or as a file store, you can begin to see the parallels to the malware, spyware and viruses that have wreaked havoc upon every operating system and storage device that there ever has been.
thus it will become important, perhaps at some later juncture, to consider the importance of fighting microspam in the microformatted ecosystem.
i don’t know how, i don’t know what that fight will necessarily look like, but i do know that it’s coming and that we ought be ready when it comes.
to this day, humans have not irradicated the common cold. nor is it likely that they’ll prevent or end the onslaught of a contagion like spam. thus as with the great potential that the lowercase semantic web brings, it is also our responsibility and charge to begin to think about how we might prevent its abuse.
Factory Joe 
There are a few separate problems here. One is that microformats could be used to mark up false information. Related is the problem that a microformat, whether maliciously or just through error, may not properly reflect the semantics of the content as humans would see it. Another is that the same item is described by a microformat in many places and so swamps aggregators. All these problems seem to be related to the lack of a way of “referring” in microformats. They are by definition about the content they are embedded in and this has some downsides.
Perhaps the latter problem is what you would term spam. And you’re right, it definately needs to be thought about since, in the ubiquitous microformat future you are imagining, this will happen all all the time quite innocently when different event sites talk about the same event for example. In the uppercase Semantic Web utopia you might show this by all referring to a common URI for the event. You can do sort of the same thing with the url element of a hEvent if it happens to have a canonical URL. With other microformats it’s not so clear how to do this.
The first two issues, of false information in microformats, stem from their “describe your own stuff” mentality that is inherited in part from the Semantic Web. In fact, they’re more committed to it than the SW since they really do have to be embedded in content and don’t have a general way for referring. I’ve thought about this a good bit and I think there’s another layer that has to be created, either in combination with microformats or subsuming their utility, in order to overcome it. What if people could mark up content created by others. It’s del.icio.us tags vs rel=tag. Which do you trust more? There’s an old Clay Shirky post that I keep coming back to whenever I think about these things.
I have some practical experience in dealing with microformats but it’s limited so I’m definately interested to hear what others have to say on the potential hurdles to the coming uF proliferation.