If your underlying reason for updating only your browser is security
Nope, it isn't.
First underlying reason is, that I have to test Websites on the latest stable release of a browser (as most of the users are windows users who update immediatley).

Second underlying reason is, that I work pretty much within the browser, and 3.5.3 (the really offical latest stable release, not the 3.5.5 one in the ppa) is much faster than 3.0.x specially concerning ajax-driven webapps.
Third underlying reason specially for ff3.5 is, that I want/have to test the new standards and functionalities (like embedding ogg directly, without a plugin).
But: This is my taste, and I can live with a manual update of my browsers, as I know how to do it

I never intended to express criticism on the release cycle and guidlines which linuxmint sticks to! For most users ff 3.0 is quite ok. For most webdevelopers, well, we can live with ff 3.0 quite good, as long as we have to live with ie6.

to go to their website and risk a phishing attack
? how should this phising attack lookalike?
I go to
http://www.mozilla.org (i do not type 'firfox' to google, and continue to the first result). As browsers are my daily bread, I know how the mozilla website should look. I am able to check md5sums.
Downloading the latest "stable" version from Mozilla wasn't actually stable at all, and I had also lost the ability to manage the package via apt/synaptic, because I had compiled it from source. In my experience with several Jaunty variants over the last few weeks, FF 3.0 is better than FF 3.5.
I use Firefox 3.5 on several boxes (linuxmint, opensuse, fedora, windows xp/vista/7) since months and I can't see any problems. Even one of my customers is running on firefox on about 15 workstations since months, and they are not complaining about unstable systems!
But I did not compile the browser myself (i trust in mozilla;) ).
The only software i am compiling myself is software like eaccelerator, php, unrtf, catdoc, ... as i have there my modified make scripts.