Oh dear. Some wise guys have discovered a JavaScript exploit in Twitter's web interface, which uses an onMouseOver instruction to hijack your own tweeting voice and force you to say things you don't want to say. Simply put, hovering on some of these colorful new tweets can result in you tweeting out the spammiest spam you ever did tweet. So, as with Tetris, be wary of those blocks of color, they are the harbingers of doom. And until the Twitter crew wrap their brains around sealing this vulnerability off, we'd recommend just using any of the cornucopia of Twitter apps floating about in the webosphere.

[Thanks to everyone who sent this in]

Update: The Twitgineers are already dealing with the issue and are rolling out a patch that should span the entire Twitterverse before too long.