There's a new binary that recommends moving to BitLocker during install, and the signature matches.

Edit: with a new, compromised key.

Project on SF is still available if you have a direct link:

http://sourceforge.net/projects/truecrypt/files/TrueCrypt/

http://sourceforge.net/projects/truecrypt/?source=navbar

http://sourceforge.net/p/truecrypt/activity/?page=0&limit=10...

Odd, 6 hours ago someone updated the TruCrypt-key.asc files, then 3 hours later posted all the new binaries.

Also odd is whoever posted the new binaries completely yanked all the previous ones, leaving only the new and questionable binary available for download.

hmm...

Looks like it's the same key as before (F0D6B1E0)

On a day like this, compare the entire key.

It's the same. Original key: http://pgp.mit.edu/pks/lookup?op=get&search=0xE3BA73CAF0D6B1...

And the new key can be found on the SF site.

Is source still available? Can we check the commit tree for anything suspicious lately? Can someone compile it and check the hash against the 7.2 binary being offered?

It's not that simple. It won't match anyway. Signatures, compiler versions, SDK versions, etc.

This guy managed to compile a previous version and have it match the released binaries https://madiba.encs.concordia.ca/~x_decarn/truecrypt-binarie...

Not quite. First, it was a ton of work to do so. Second, in the end he in fact only managed to match most of the released binary. Then, for most of the unmatched parts he came up with some solid arguments why they would be different (timestamps, pathnames and such). However, in the end there were still a couple of chunks of unmatched data that he couldn't explain why it has the value it does. Finally he argues that these chunks probably can't contain any hidden backdoors. This last bit I have a bit of a problem with because it's just speculation, all the rest of the research is very solid and without guesswork, and the conclusion of the article IMO incorrectly states he "has matched the binaries".

I'm chalking this up to, after all that work, having come so far, kind of wanting to say you did it, and providing a few (IMO) hand-wavy arguments why those last impossible chunks of bits probably don't matter, instead of having to admit defeat after being able to account for 99.9% of all bytes.

I can understand that, I've been wanting to type "but it's probably not backdoored" three times already typing this post, and catching myself because really I have nothing to go on to make or back that statement.

Yes, but did they sign it using the same key they were using before?

edit: Apparently not, according to the link @Alupis posted.