Thanks, I can go back to my friends now and tell them my computer is inhackable

But see also: http://en.wikipedia.org/wiki/MD5

Specifically the section titled "Security" and "Collision vulnerabilities."

The first sentence of the Security section:
md5sum produces MD5 hashes. Therefore, if a flaw exists in the MD5 algorithm, md5sum will be susceptible to an exploitation of the flaw.

Yes, it's not only possible but also relatively easy. If you google about MD5 collisions you'll have tons of material to study. I remember a while back there was this guy/group that posted the MD5 of a PDF with the results of a future election on them, and when the election took place they produced the respective PDF (which matched the previously-published MD5) and they were right. Or perhaps the whole election thing was just for illustrative purposes, I don't remember well. The point being, of course, that they had PDFs with different candidates on them and the PDFs all had the same MD5 message digest. IIRC, they were quite open to explaining how they did it, and there was source code available for anyone wishing to replicate their experiments (if not, there was certainly source code for a toolkit somewere else, as I know I saw it).

In any case, letting you know that this information is publicly available is pretty much as far as we can go here. Any member offering you specific help with this would be in violation of the LQ Rules. You're new here, so it's understandable that you weren't aware of this yet, but you are now.

I'm not sure that Tripwire would be an option, as you wouldn't be able to get a known-good baseline.

That said, one of the nice things about HIDS like Tripwire is that they use multiple hash functions (MD5, SHA-1, etc.), so sneaking a file modification past a scan is terribly difficult (program bugs aside), since you'd need to do it in a way that fools every single function. This remains true even if one of the functions is broken (such as MD5, for example).