Keep in mind that you yourself only just got the ability to add a community tag. Bots will not have an easy time gaining this ability. Vindictive users are usually vindictive in all aspects of their use and easily identified and filtered.
Filtering is not an option in this case, though.
So let's say I decide to create a bot or even just a separate account to screw with people's tags. All I have to do is post a bit and get one hubwheel (like I just did), and then I can use this account forever.
Because you are saying using filtering is how you will prevent people from doing this, that is on a user level. Users don't know what community tags are suggested or voted on at all, so they won't know who to filter if the vandals just compartmentalize multiple accounts.
So I can go right now into #blog for instance and just strip off everyone's community tag and randomize the tags to be completely off-topic, and now the only valid tag they have is their #blog tag. The only way to trace who did this would be for admins to get involved, which I was under the impression that your designs were to give the users control over moderation.
Yes, the community can go in and fix all that, but if it were an automated process or just something someone spent a little bit of time on each day, would you want all your users to do this? They'd have to fix every single post every single time. Plus, it wouldn't be that tag that the poster chose, it would be exclusively a community tag at that point (hence, why I said you essentially got rid of the second tag).
Do you as full-time job possessing admins also want to go around policing the site to figure out who these users are?
This is mostly vandalism and juvenile, but we're talking about internet users here :)