Schneier, UW Team Show Flaw In TrueCrypt Deniability

An anonymous reader writes "Bruce Schneier and colleagues from the University of Washington have figured out a way to break the deniability of TrueCrypt 5.1a's hidden files. What about the spanking-new TrueCrypt 6? Schneier says that 'The new version will definitely close some of the leakages, but it's unlikely that it closed all of them.' Meanwhile, PC World is reporting that the problems Schneier and colleagues found are bigger than just TrueCrypt. Among their discoveries: Word auto-saves the contents of encrypted files to the unencrypted portions of your disk, and this problem should apply to all non-full disk encryption software. Their research paper will appear at Usenix HotSec '08."

usenix what?

HotSex 08? Where do I sign up!

Lucky for me...

I encrypt using a one way algorithm know as "fire" that transforms all my secrets into ashes.

Since matter can not be destroyed, only changed, decryption is just around the corner. Also, AJAX will be used somehow.

Bay area venture capital welcome!

Sorry, dude...

Seems that someone found a semi-reliable decryption mechanism that can not only stand up to that, but can reverse an even stronger algorithm known as "volcano" [byu.edu].

Didn't mean to dash your dreams, but you know how the security game goes...

/P

Re:Sorry, dude...

"Volcano" is, indeed, a stronger algorithm than "fire", but it's also much coarser-grained. Further research shows that the decrypted portions were not completely encrypted, merely provided with a partially-encrypted wrapper.

We can also discuss the even more advanced "Thermonuclear ground-zero" algorithm, but the ultimate form of this type of encryption (matter-antimatter annihilation) is only theoretically possible with our current technology.

Re:Sorry, dude...

Thermonuclear ground-zero encryption is unnecessary, you just need good a good Brownian crypto device.

On a serious note, there's also steganography. I wrote up a tool that works like shred(1), except instead of DoD-compliant type over-writes, it uses blocks of harmless text from Project Gutenberg. Theoretically it's weaker than a 35-pass algorithm, but the advantage is that it's now much harder to retrieve the original data, since it's much harder to tell apart.

I really want to do something that would get my computer seized by the NSA so I can laugh while imagining them trying to find the data they're looking for. "Aha! I've found some unencrypted text... it says, 'Of all the cants which are canted in this canting world, â" though the cant of hypocrites may be the worst, â" the cant of criticism is the most tormenting...' Never mind, it's just some crap again...."

Anyone know how to get in touch with Osama bin Laden?

Re:Lucky for me...

I encrypt using a one way algorithm know as "fire" that transforms all my secrets into ashes.

Is that the algorithm invented by the Greek hacker, Prometheus? I heard he got in a bit of trouble over it, he ended up somewhere like Guantanamo, but eventually was rescued.

Word and what?

Among their discoveries: Word and auto-saves the contents of encrypted files to the unencrypted portions of your disk,...

If you're like me (meaning that you pay attention to what you read), you may be wondering what in the world "Word and auto-saves" means. I wondered so much I even followed the link, and saw that the omitted term was Google Desktop, omitted because of very sloppy cut and paste of the article.

About Bruce Schneier

Some of you may not be aware of the stature of Bruce Schneier in the field of computer security, so here is some background information:

http://geekz.co.uk/schneierfacts/facts/top [geekz.co.uk]

Bruce Schneier once decrypted a box of AlphaBits.

Most people use passwords. Some people use passphrases. Bruce Schneier uses an epic passpoem, detailing the life and works of seven mythical Norse heroes.

Bruce Schneier knows Alice and Bob's shared secret.

Vs lbh nfxrq Oehpr Fpuarvre gb qrpelcg guvf, ur'q pehfu lbhe fxhyy jvgu uvf ynhtu.

Bruce Schneier's secure handshake is so strong, you won't be able to exchange keys with anyone else for days.

Bruce Schneier knows the state of schroedinger's cat

Bruce Schneier writes his books and essays by generating random alphanumeric text of an appropriate length and then decrypting it.

When Bruce Schneier observes a quantum particle, it remains in the same state until he has finished observing it.

If we built a Dyson sphere around Bruce Schneier and captured all of his energy for 2 months, without any loss, we could power an ideal computer running at 3.2 degrees K to count up to 2^256. This strongly implies that not only can Bruce Schneier brute-force attack 256-bit keys, but that he is built of something other than matter and occupies something other than space.

Though a superhero, Bruce Schneier disdanes the use of a mask or secret identity as 'security through obscurity'.

Re:About Bruce Schneier

Personally, I like "Bruce Schneier already has a backup plan for when the second person discovers P=NP."

Re:About Bruce Schneier

I ran into Bruce Schneier at an airport once. While we were waiting for a plane, I asked him if he would show me a "cool computer trick". He popped the RAM out of my laptop and quickly tasted the edge with the gold leads. He then told me that at 11:23pm the previous night I had visited ideepthroat.com with Firefox. Damn he's good.

Summary is inaccurate

Schneier et al don't break TrueCrypt's deniability, per se. They simply show that Word, Google Desktop, and other automatically-indexing programs may reveal a hidden partition's possible existence.

This is a concern, of course, but can be avoided by careful use of the software invoked when using a TrueCrypt partition (i.e. killing processes except for TrueCrypt, etc).

I believe there's also a portable version of TrueCrypt that can be used that leaves no traces on the OS install once you're finished.

Deniability on SSD?

This has been bugging me and I wonder if anyone out there can answer this: would the write-leveling used by flash drives defeat deniability as well? After all, if the most recently written-to portions of the drive are in a supposedly unused block, isn't that a bit of a giveaway?

Re:Deniability on SSD?

the Truecrypt documentation mentions the possible implications of this.

Wear-Leveling

Some storage devices (e.g., some USB flash drives) and some file systems utilize so-called wear-leveling mechanisms to extend the lifetime of the storage device or medium. These mechanisms ensure that even if an application repeatedly writes data to the same logical sector, the data is distributed evenly across the medium (logical sectors are remapped to different physical sectors). Therefore, multiple "versions" of a single sector may be available to an attacker. This may have various security implications. For instance, when you change a volume password/keyfile(s), the volume header is, under normal conditions, overwritten with a re-encrypted version of the header. However, when the volume resides on a device that utilizes a wear-leveling mechanism, TrueCrypt cannot ensure that the older header is really overwritten. If an adversary found the old volume header (which was to be overwritten) on the device, he could use it to mount the volume using an old compromised password (and/or using compromised keyfiles that were necessary to mount the volume before the volume header was re-encrypted). Due to security reasons, we recommend that TrueCrypt volumes are not stored on devices (or in file systems) that utilize a wear-leveling mechanism. If you decide not to follow this recommendation and you intend to use system encryption when the system drive utilizes wear-leveling mechanisms, make sure the system partition/drive does not contain any sensitive data before you fully encrypt it (TrueCrypt cannot reliably perform secure in-place encryption of existing data on such a drive; however, after the system partition/drive has been fully encrypted, any new data that will be saved to it will be reliably encrypted on the fly). To find out whether a device utilizes a wear-leveling mechanism, please refer to documentation supplied with the device or contact the vendor/manufacturer.

I said it before, I'll say it again

Windows caches all types of stuff about filesystems it touches in the registry. Open regedit some time and search for "OpenSaveMRU" and you'll see that pretty much every file you click to open in Windows is in there.

Not that Linux is any better, at least Gnome systems - check out ".nautilus" in your home folder. Same thing going on there with the directory structure, you name it. The first thing I do on a new Ubuntu box is remove ".recently-used.xbel" and create a directory with the same name, and make ".nautilus" owned by root and not world-writable. /tmp is obviously a problem on Unix-type systems as well, along with the swap partition.

Of course if your whole system is encrypted these are not problems, but then you don't exactly have a deniably-encrypted filesystem.

Re:Get A Mac

Windows should build in a encryption program like on Mac OS X

Uh... they did... 8 years ago.

They've had EFS (encrypting file system) since Windows 2000.
http://en.wikipedia.org/wiki/Encrypting_File_System [wikipedia.org]

They've added BitLocker Drive Encryption with Vista (Ultimate & Enterprise).
http://en.wikipedia.org/wiki/BitLocker_Drive_Encryption [wikipedia.org]

Re:Get A Mac

My bet would be that if you have the DFS filesystem mounted, then Spotlights (or Beagle on Linux) would just index it like any part of the filesystem.

They're not trying to decrypt files here, but just prove that files exist. TrueCrypt lets you put an encrypted volume inside an encrypted volume, such that if you mount the "outer" volume, you can't show evidence that there even exists an "inner" volume. However, if you mount that "inner" volume and use the files in it, Windows will make a Recent Documents shortcut to its location, thus disclosing the fact that there are files there.

I'm a TrueCrypt user, but not a DFS user, since I care more about the encryption than I do about plausible deniability, but I'm interested in trying this out. The test case might be along the lines of:

• Mount a DFS volume on a Mac
• Do a spotlights search for something inside that volume
• Unmount the DFS volume
• See if theres any cached data from Spotlights that still hints at the existence of the file within your hidden filesystem

Since Spotlights also does a full-text search, does it cache any of that full-text data to make the next search faster?

Re:Get A Mac

Spotlight's index is stored in the root of the volume it's indexing. Encrypted filesystems are independent volumes, so their indexes are stored in their volume root. The index of the primary filesystem isn't altered.

I'm not sure it leaks zero information -- there have been some bugs with Spotlight indexes and FileVault-encrypted home directories.

Re:Get A Mac

Use this l33t HaXX0r tool called regedit?

User Key: [HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Policies\ Explorer]
System Key: [HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\Policies\ Explorer]
Value Name: NoRecentDocsHistory
Data Type: REG_DWORD (DWORD Value)
Value Data: (0 = disable restriction, 1 = enable restriction)

However, if you mount that "inner" volume and use the files in it, Windows will make a Recent Documents shortcut to its location, thus disclosing the fact that there are files there. I'm a TrueCrypt user, but not a DFS user, since I care more about the encryption than I do about plausible deniability, but I'm interested in trying this out. The test case might be along the lines of:

Re:Get A Mac

Really?

All of Mac OS X encryption operates on user-managed encrypted disk images (volumes) or "encrypted home directories" (FileVault), which is really an OS-managed encrypted disk image.

FileVault home directories are no stronger than your login password. As this password is stored hashed only once (albeit salted, as of 10.4), it had better be immune to brute-force-guessing. They're also only as strong as your system-wide FileVault recovery keychain, as a copy of the key is stored in that, too.

Non-FileVault encrypted images at least use 1000-round PBKDF rather than a single hash and don't, by default, use a recovery keychain. At only 1k rounds, though, it had still better be immune to brute-force guessing.

None of this addresses the fact that using a Mac OS X system with an encrypted directory still leaks information about the contents of that directory onto the unencrypted parts of the drive. In fact, if anything, TrueCrypt is better about not doing this than the Mac, though neither of them hide their tracks all that well. The best approach is to have TrueCrypt running full-disk encryption so that there's nowhere for data to leak to.

Re: BitLocker Backdoor- Source?

I'm replying to myself, but I have additional info to add.

[...] it captures live data on the computer, which is why it's important for agents not to shut down the computer first, Fung said. A law enforcement agent connects the USB drive to a computer at the scene of a crime and it takes a snapshot of important information on the computer. It can save information such as what user was logged on and for how long and what files were running at that time, Fung said. It can be used on a computer using any type of encryption software, not just BitLocker.

So it looks like COFEE is a USB device that performs monitoring once Vista has been booted and logged in. Not having your BitLocker USB drive plugged in and not leaving your PC on would seem to defeat an attack by COFEE.

Re:Get A Mac

So, just to play along, what software do you propose to use on the mac to provide deniable encryption?

You could try this program called TrueCrypt [truecrypt.org]. It seems to work okay.

Re:Let me get this straight

Anyway, now Im rambling, but I use truecrypt only on my secure linux box, which doesnt have these problems

Are you sure? Have you checked your ~/.bash_history file? Are you sure your editor isn't leaving autosaves in /tmp? There could even be plain text in your swap partition. It's hard to really know.

If I needed plausible deniability I'd put a virtualbox image in the deniable container. Then I'd turn off swap and link ~/.bash_history to /dev/null. And I'm sure I've forgotten something.

Re:Let me get this straight

It seems to me that the best way to get this done would be for a bunch of guys (ideally with the paranoia of the OpenBSD guys) set about creating a Linux distro with all these things built in. It would obviously not be one built for performance, but it would be fully secured out of the box with encrypted swap, /tmp set as a ramdisk (optionally for users with enough ram or encrypted for those who don't), all installed apps (from vim to OpenOffice) configured to use secure areas for temp files etc etc.

Such a distro would mean having that level of paranoia would not arouse as much suspicion, as you could just say "Meh, I run Paranoia Linux coz I heard it was secure" and not look like you put much effort into it.

So, any takers on this project? I would, but I'm sucky at this kind of thing.

Re:And this is exactly why..

you run at least full disk encryption. If one needs further plausible deniability, THEN you can run truecrypt. Also, cleaning out temp files should be a regular occurrence, as should running on an encrypted swap file/partition.

This is why secutiry needs to be left to the professionals and requires scrutiny. It is very hard to get right and very easy to leave holes. You run full disk encryption, but in many parts of the world, you can be compelled to disclose your keys. So, since your keys are disclosed, you now may as well assume that you never had the encryption in the first place. That puts you right back to square 1 and there is now evidence that you have a hidden volume.

Full disk encryption protects you against the consequences of theft, and for this, deniability has no utility. Deniability protects you against certain governments, and for this, full disk encryption often provides little utility.

Re:No Problem Here

"Keep in mind, though, that you can simply add exceptions to your updatedb.conf file, such that the directories/partitions you list will not be indexed (and hence will not be locatable by slocate)."

yes, put your hidden directories/partitions in /etc/slocate then slocate will not reveal their existence.

It seems to me there is something wrong with this sheme but I cannot put my finger on it. Hum ... but then again I'm not a security specialist.