Why not have a VM just for internet banking? Surely that's easier than taking the chop out of the safe and waiting in line at the branch.
I wouldn't trust myself to keep the latest version of Windows secure, given that I haven't used Windows in over a decade, so I'm pretty sure I can't be trusted to keep XP safe. (As far as I'm aware, the "security" software that the bank's ActiveX control communicates with only runs on XP.)
It's simple, really.
Set up the VM. Do a snapshot. Every time you need to use the shit website, launch the VM, open the site in IE and nothing else. When you're done, shut down the VM and reset it to the known good snapshot.
There. Secure.
for example by enumerating the connected PCI devices and looking for common VM vendors virtual devices.
>Doesn't that defeat the purpose of the VM to begin with
that depends on your use-case. If it's about separating mostly trusted applications and/or servers, then absolutely not.
If it's about investigating known-bad code, then, yes, absolutely - malware is often intentionally disabling itself when it detects it's running in a VM.