Suppose that I have access to a computer (say, running OSX, so that we have BSD tools available) on which I have an admin-privileged account (full superuser rights), but that the owner of the computer has, for contractual/legal reasons, installed some kind of security program that can log keystrokes and probably other spyware-type things. As a matter of the computer owner's policy, I'm allowed to use the computer for some personal activity, but I am not permitted to disable or circumvent the security software. (edit: Booting the computer to a state without the security software running is not permitted. I am also not permitted to use my own computer.)
What ways are there for me to access my personal accounts of various types (general email, Google accounts including Google Music, Pandora, servers via SSH, etc.) without exposing the accounts to potential compromise from the information logged by the security software?
One possibility that has occurred to me is that logging in to a service that used 2- or 3-factor authentication might be safe, especially if some of the factors were out-of-band. For instance, if I logged in to an OpenID provider such as MyOpenID by using a username, password, and CallVerifID/PhoneFactor, my username and password would be captured by the security software, but the account could not be accessed without the something-I-have factor of my phone, used entirely out-of-band. (I think similar reasoning would hold for an OpenID provider using a one-time-password token such as a Yubikey, even though the OTP isn't sent out-of-band.) I would then think that any site I then logged into using that OpenID account would still be secured. Am I right about this?